Autonomic nervous system and endocrine system response to upper and lower cervical spine mobilization in healthy male adults: a randomized crossover trial

ABSTRACT

Background

Cervical spine mobilizations may differentially modulate both components of the stress response, consisting of the autonomic nervous system and hypothalamic pituitary adrenal-axis, depending on whether the target location is the upper or lower cervical spine. To date, no study has investigated this.

Methods

A randomized, crossover trial investigated the effects of upper versus lower cervical mobilization on both components of the stress response simultaneously. The primary outcome was salivary cortisol (sCOR) concentration. The secondary outcome was heart rate variability measured with a smartphone application. Twenty healthy males, aged 21–35, were included. Participants were randomly assigned to block-AB (upper then lower cervical mobilization, n = 10) or block-BA (lower than upper cervical mobilization, n = 10), separated by a one-week washout period. All interventions were performed in the same room (University clinic) under controlled conditions. Statistical analyses were performed with a Friedman’s Two-Way ANOVA and Wilcoxon Signed Rank Test.

Results

Within groups, sCOR concentration reduced thirty-minutes following lower cervical mobilization (p = 0.049). Between groups, sCOR concentration was different at thirty-minutes following the intervention (p = 0.018).

Conclusion

There was a statistically significant reduction in sCOR concentration following lower cervical spine mobilization, and between-group difference, 30 min following the intervention. This indicates that mobilizations applied to separate target locations within the cervical spine can differentially modulate the stress response.

KEYWORDS:

• Manual therapy
• musculoskeletal manipulations
• heart rate variability
• hypothalamic pituitary adrenal-axis
• cortisol
• stress response

Introduction

The body’s stress response has both central and peripheral components [1–3]. The peripheral components consist of two interrelated systems: the autonomic nervous system (ANS) and the hypothalamic pituitary adrenal-axis (HPA-axis) [1–3]. For the purpose of this study, ‘stress response’, refers to the peripheral components. Once exposed to a stressor, be that physical, physiological, or psychological, homeostasis of the human body is threatened, and the stress response functions to return the body to equilibrium [4]. The stress response initially activates the fast acting, but short lasting ANS response, which then communicates with the HPA-axis for the slower acting but longer lasting endocrine response [1,4–6]. Communication between the ANS and HPA-axis is interchangeable and bidirectional [7]. Consequently, the stress response either heightens or dampens its response [1,4–6]. A heightened response may present with an increased heart rate and hyperarousal, whereas a dampened response may present with a reduced heart rate and hypoarousal [8].

Manual therapy is a common treatment modality used in the physiotherapy, osteopathy, and chiropractic professions. Despite demonstrating clinical efficacy and effectiveness [9–11], the underlying mechanisms of spinal manual therapy are not fully understood [12,13]. One hypothesis is that manual therapy may exert effects on the ‘stress axis’ or ‘stress response’ [14]. Manual therapy applied to the cervical spine has been shown to modulate isolated components of the stress response; either the ANS [15–22] or the HPA-axis [23,24]. However, manual therapy targeting the thoracic spine [25,26] or craniosacral region [27] has been shown to modulate both components of the stress response in unison. Evidence also suggests that manual therapy applied to distinct anatomical levels within the cervical spine can elicit a differential ANS [28–30] or HPA-axis [24] reaction. Currently, no study has investigated the physiological effects of cervical spine mobilization on both components of the stress response simultaneously, depending on the segments targeted.

A commonly used way to measure the HPA-axis response is by measuring cortisol, a stress hormone released from the adrenal cortex [31]. Salivary cortisol (sCOR) measures represent a noninvasive, convenient, and reliable way to measure HPA-axis activity [32]. HRV is a proxy measure of the cardiac ANS, more specifically the parasympathetic nervous system (PSNS) modulation on the heart, otherwise known as vagal tone [33–35]. Root mean squared of successive differences of RR intervals (rMSSD) is the chosen HRV metric for this study as it has been shown to be a reliable indicator of PSNS activity [36].

The aim of this work was to explore the physiological effects of mobilization to the upper versus lower cervical spine in healthy males, by comparing: (1) HPA-axis response measured with sCOR, and (2) the ANS response measured with HRV. This study will provide a deeper understanding of the mechanisms of cervical spine mobilization and has important implications for clinicians, whom may be able to selectively modulate the stress response depending on their target location within the cervical spine.

Methods

Study design

A randomized, crossover, repeated measures trial was conducted to evaluate the time-course effects of upper versus lower cervical mobilization on sCOR concentration and HRV. A crossover design was selected to minimize potential confounding by individuals’ specific covariates as each individual acts as their own control [37]. The study was reported in accordance with the CONSORT 2010 statement: extension to randomized crossover trials [37] (Appendix 1), and TIDieR checklist (Appendix 2). Ethical approval was obtained from the University of Otago Human Ethics Committee (H21/086). A protocol was registered a priori with Australia New Zealand Clinical Trials Registry (ACTRN12621001035819). All participants signed informed consent prior to enrollment in the study.

Participants

Healthy males, aged between 21 and 35 were recruited from the local community. The age range was selected due to maturation of the ANS before the age of 20 [38], and decline of the endocrine system after the age of 35 [39]. Participants were required to have a smart phone. Participants were excluded if they were taking systemic glucocorticoid or cardioactive medication; had an endocrine, central nervous system, cardiovascular, mental health, co-morbid musculoskeletal, or history of serious pathologic or psychiatric disorder. Participants were also excluded if they had any contraindications, and on a case-by-case basis if they had any precautions, to manual therapy [40]. Females were excluded due to the differential effect of sex [41,42], and the menstrual cycle [43–45], on our outcomes of interest.

Intervention

A mobilization technique was performed on all participants. These techniques consisted of low frequency, mid-range, bilateral, and alternating oscillations of either the upper cervical spine (C0–1 and C1–2) or lower cervical spine (C6–7 and C7-T1). Low frequencies have been associated with ANS processing and mobilizations can be performed within this frequency range [46,47]. Therefore, a frequency of 3 Hz was selected for each mobilization [47]. Each level was mobilized for three sets of two-minutes, separated by a 30-s rest period. All interventions were performed by the same operator, a certified musculoskeletal physiotherapist specialist with over 30 years clinical experience. To control for potential contextual influences, each intervention was performed in the same room at the School of Physiotherapy, University of Otago [48], at the same time [8,49–51] for each participant, between 1300 and 1700. Temperature was accounted for, and no natural light was allowed to enter the room.

Outcome measures

Salivary cortisol

Cortisol is a hormone released into the blood stream by the adrenal cortex and is the end product of the HPA-axis and a proxy measure of HPA-axis activity [52,53]. Whole saliva samples were collected with SalivaBio’s 2 mL cryovials and Saliva Collection Aids (Salimetrics, State College, PA) via the unstimulated passive drool technique [32] (Appendix 3). This technique is gold standard and is also noninvasive and the most reliable method for collecting sCOR samples [32]. Our primary outcome was the sCOR concentration at 30-min post-intervention. Concentrations of cortisol peak within the blood 10–30 min following a stressor, and then peak in the saliva 2–3 min following saturation in the blood [54].

Heart rate variability

Heart rate variability:

HRV is the beat-to-beat variation of R-R intervals from the cardiac cycle [35]. HRV provides a noninvasive measure of cardiac ANS activity, predominately vagal tone [35,55]. An iPhone application, Camera HRV (A.S.M.A. B.V.), was used to measure HRV with a technique called photoplethysmography (PPG). PPG uses the smartphones light to illuminate the skin and detect blood volume variation during a cardiac cycle [56]. PPG is a reliable way to measure HRV and is validated when compared to electrocardiogram, the gold standard of HRV collection [57–59]. Many metrics exist to quantify HRV [34,35]. For this study, the rMSSD was chosen as the primary time domain metric to quantify vagal tone [36]. rMSSD measured with PPG has a low standard error [60]. Please refer to appendix 4 for factors accounted for which can affect the outcomes.

Outcome measures data collection

Salivary cortisol

Participants collected sCOR samples at 14 timepoints. Eight were collected at home, and six were collected in clinic (Figure 1). Cortisol exhibits a diurnal decline, with peak concentrations in the morning 30–45 min after awakening and a gradual decline over the course of the day, reaching a minimum around midnight [61–63]. Baseline cortisol samples were collected to [1]; plot the diurnal decline of cortisol concentrations [2,61] assess the accuracy of measurements with baseline and age-matched normative values [51,64]. Midnight sCOR samples have the highest reliability when assessing chronic HPA-axis tone [65]. Hence, sCOR was collected the night after each intervention (post-pm) to assess the response of sCOR to the intervention. sCOR was collected in the seated position.

Figure 1. Timing of outcome measurement. During each intervention, sCOR concentrations and HRV data were collected in clinic at three timepoints, baseline or five-minutes prior to the intervention (pre-5), then five-minutes (post-5) and thirty-minutes (post-30) following the intervention. Baseline sCOR samples were collected at home on day five and six between the hours of 0600–0800 (baseline-am), 1400–1600 (baseline-afternoon), and 2200–0000 (baseline-pm). sCOR samples were then collected the night following each intervention between the hours of 2200–0000 (post-pm). HRV data was collected each morning on waking, at home, between the hours of 0600–0800. Day two to seven, and day nine to 14, were baseline HRV data (baseline-am). Day eight and 15 were post-intervention HRV data (post-am).

Heart rate variability

Participants collected HRV data at 21 different timepoints. Fifteen were collected at home, and six were collected in the clinic (Figure 1). Baseline measurements were collected over six consecutive days prior to each intervention [66,67]. The day one measurement was omitted from analysis as a practice measurement. The gold standard for HRV measurement is in the middle of the night while the individual is sleeping, as this is when vagal tone is at its highest [68]. This timing is not possible when using a smartphone application, hence our measurements were taken first thing in the morning upon waking, when the participant should still be in a vagally dominant state [49,50]. Participants were asked to breath spontaneously [34], and measure their HRV for two-minutes [34,69]. All HRV measures were taken in supine [34,58], prior to sCOR collection, to minimize the autonomic effect of position change on HRV [70].

Salivary cortisol data analysis

sCOR samples were analyzed with a Salimetrics ELISA immunoassay kit by a physiologist at a University of Otago physiology laboratory in accordance with the standard procedures (Appendix 5).

HRV data analysis

Raw RR intervals were exported, deidentified, stored, processed, and cleaned through Kubios HRV Standard version 3.5.0 (Biosignal Analysis and Medical Imaging Group, Kuopio, Finland) (Appendix 6).

Procedure

Potential participants attended an enrollment appointment and were screened against the inclusion and exclusion criteria. Included participants were provided with the saliva collection kit and smart phone application, along with detailed verbal and written instructions. Participants were requested to fill out an individual diary, accounting for various factors which can influence HPA-axis and ANS activity (Appendix 4). Perceived stress influences the HPA-axis and ANS [6,71]. To account for this, participants were requested to fill out the perceived stress scale (PSS-10) on day 1, 7, and 14. The PSS-10 has high reliability and validity, with higher scores indicating higher levels of perceived stress [72].

Following baseline measurements, participants’ interventions were separated by a conservative one-week washout (Figure 1). Considering our study’s healthy population, and comparable studies reporting a response no greater than minutes to hours after manual therapy techniques [13,24,26,73], a one-week washout period was ample. Upon arrival, participants were screened for any change in health status, or adverse reaction following the previous intervention. For environmental acclimatization, participants lay supine with extended legs, and avoided moving, speaking, or using their phone, for 5 min prior to pre-intervention measures [34,74]. The intervention was performed on the target location, followed by post-intervention measures.

Randomization and blinding

To prospectively code saliva collection vials, each participant selected a sealed opaque envelope containing their participation number upon enrollment. A research assistant, independent to the outcome assessor, used a computerized random number generator to develop a 1:1 ratio randomization table linking the participation number with their intervention block. It is not possible to blind the treating clinician [75] and is unlikely the participants were blinded to which intervention they received. The outcome assessor only analyzed de-identified data.

Power calculations

We are unaware of any studies investigating upper or lower cervical mobilizations on our primary outcome, and the minimally clinically important difference for cortisol is currently unknown [76]. The target location of mobilizations are thought to be anatomically separate, therefore we anticipated a large effect size of d = 0.8 [77]. The upper cervical spine is in close proximity to the brainstem and PSNS [78,79], whereas the lower cervical spine is in close anatomical proximity to the inferior cervical (or stellate) ganglion, which are paravertebral ganglia of the sympathetic nervous system [80]. An a-priori sample size calculation was performed using G*Power (Version 3.1.9.7). A sample size of n = 15 was required to provide 0.8 power at an a = 0.05 with a two-tailed test for a repeated measures crossover trial. Estimating a 20% drop out, we require 20 participants for even samples of 10 in each group.

Statistical analyses

Participant characteristics and outcome variables were presented with descriptive statistics. Data normality and skewness were checked by visual inspection of histograms, and formally with the Shapiro–Wilk test [81]. Data and residuals were non-normally distributed, despite data transformation and outlier correction [82]. Outliers were assessed through box and whisker plots and were winsorized if correlated with a confounding factor from participants clinical diary [83]. Non-parametric tests were used for our non-normally distributed data [84], from a small sample size [85]. Within the clinic, a Friedman’s two-way ANOVA tested the within-group factor (time) on the outcomes [86]. A Bonferroni correction adjusted for multiple post-hoc comparisons [86] and accounted for a type 1 error [87]. The Wilcoxon-signed rank test was used for the post hoc, between-group, pairwise comparison [87]. A pairwise comparison was performed with the Wilcoxon-Signed Rank Test for the within-group measurements taken at home [88]. Appropriate statistical analyses were performed to rule out any carryover, sequence, or period effects [89]. Effect size was calculated with the formula r = z/√N [87]. Missing data was minimal (3.5% for sCOR, and 2.8% for rMSSD) and was handled with pairwise deletion [90]. Data analysis was performed using IBM SPSS Statistics version 28.0 (IBM Corporation, Armonk, NY).

Results

The flow of participants is presented in Figure 2 [37]. Data collection occurred between the 13^th of September and 13^th of October 2021. Twenty males participated in this study. Participant characteristics are described in Table 1. No participants were lost from this study. Two HRV datasets were excluded due to technical and operator error. Eleven individual measures were lost due to sample contamination, operator and clinician error. No adverse reactions were reported following any manual therapy interventions.

Figure 2. Flow of participants.

Table 1. Baseline participant characteristics (n = 20).

Characteristics
Number of males n = (%)	20 (100%)
Age (years)	22.85 (2.134)
Ethnicity n = (%)
New Zealand European	12 (60%)
Māori	2 (10%)
Chinese	1 (5%)
Indian	2 (10%)
Other	3 (15%)
Perceived stress scale scores
Day 1	11.500 (5.643)
Day 7	11.450 (5.916)
Day 14	12.450 (6.755)

SD: standard deviation, results are presented as mean (SD) unless otherwise stated.

Salivary cortisol

Clinic

Within groups, there was a statistically significant change in sCOR levels following lower cervical mobilization, X² [2] = 6.303. A Bonferroni correction revealed a statistically significant reduction in sCOR levels from pre-5 (Md = 0.194) to post-30 (Md = 0.194). There was no statistically significant change in sCOR levels following upper cervical mobilization, X² [2] = 4.900.

Between groups, there was a statistically significant difference in sCOR levels between the upper and lower cervical mobilization groups at post-30, z = 2.374. The median score of sCOR was lower following upper cervical (Md = 0.163) compared to lower cervical mobilization (Md = 0.194). There was no statistically significant difference in sCOR levels following upper cervical and lower cervical mobilization at pre-5, z = 1.847, or post-5, z = 1.254 (Table 2, Figure 3).

Figure 3. Clinic measurements of salivary cortisol and rMSSD.

Figures include 95% confidence interval error bars

Table 2. Clinic measurements of salivary cortisol and rMSSD.

		Pre-5	Post-5	Post-30	Within group pre-5 - post-5	Within group pre-5 - post-30	Within group post-5 - post-30
Salivary cortisol (mean ± SD) (µg/dL)	Upper	0.186 ± 0.046	0.190 ± 0.081	0.174 ± 0.054	p = 0.805 r = 0.175 n = 20	p = 0.081 r = 0.350 n = 20	p = 0.805 r = 0.175 n = 20
	Lower	0.252 ± 0.147	0.223 ± 0.092	0.215 ± 0.070	p = 0.259 r = 0.294 n = 17	p = 0.049* r = 0.412 n = 17	p = 1.000 r = 0.118 n = 17
	Between group	p = 0.065 r = 0.317 n = 17	p = 0.210 r = 0.209 n = 18	p = 0.018* r = 0.385 n = 19
rMSSD (mean ± SD) (ms)	Upper	52.589 ± 22.320	58.933 ± 26.732	48.697 ± 18.079	p = 0.259 r = 0.294 n = 17	p = 1.000 r = 0.029 n = 17	p = 0.178 r = 0.323 n = 17
	Lower	53.905 ± 22.945	51.354 ± 21.106	57.315 ± 25.787	p = 1.000 r = 0.059 n = 17	p = 0.690 r = 0.206 n = 17	p = 1.000 r = 0.147 n = 17
	Between group	p = 0.756 r = 0.055 n = 16	p = 0.163 r = 0.028 n = 17	p = 0.906 r = 0.020 n = 17

Home night

Within groups, there was no statistically significant change in sCOR levels from baseline-pm to post-pm following upper cervical, z = 0.747, or lower cervical mobilization, z = 1.067.

Between groups, there was no statistically significant difference in sCOR levels post-pm following upper cervical and lower cervical mobilization, z = 0.849 (Table 3, Figure 4).

Figure 4. Home measurements of salivary cortisol and rMSSD.

Figures include 95% confidence interval error bars

Table 3. Home measurements of salivary cortisol and rMSSD.

		Baseline	Post-intervention†	Within group
Salivary cortisol (mean ± SD) (µg/dL)	Upper	0.101 ± 0.033	0.133 ± 0.116	p = 0.455 r = 0.118 n = 20
	Lower	0.101 ± 0.033	0.174 ± 0.192	p = 0.286 r = 0.173 n = 19
	Between group	NA	p = 0.396 r = 0.138 n = 19
rMSSD (mean ± SD) (ms)	Upper	73.525 ± 25.490	65.774 ± 31.747	p = 0.048* r = 0.330 n = 18
	Lower	71.943 ± 23.213	65.072 ± 29.128	p = 0.492 r = 0.118 n = 17
	Significance and effect size between group	p = 0.647 r = 0.076 n = 18	p = 0.687 r = 0.069 n = 17

µg/dL: microgram per deciliter, ms: milliseconds, n: population, p: adjusted significance, r: effect size, SD: standard deviation, * statistically significant p < 0.05, † salivary cortisol: post-pm and rMSSD: post-am.

Heart rate variability: rMSSD

Clinic

Within groups, there was no statistically significant change in rMSSD following upper cervical, X² [2] = 4.353, or lower cervical mobilization, X² [2] = 1.529.

Between groups, there was no statistically significant difference in rMSSD following upper cervical and lower cervical mobilization at pre-5, z = −0.310, post-5, z = −1.396, or post-30, z = 0.118 (Table 2, Figure 3).

Home morning

Within groups, there was a statistically significant change in rMSSD from baseline-am to post-am following upper cervical mobilization, z = −1.982. The median score of rMSSD reduced from baseline-am (Md = 70.398) to post-am (Md = 61.183). No statistically significant change was reported following lower cervical mobilization, z = −0.686.

Between groups, there was no statistically significant difference in rMSSD post-am following upper cervical and lower cervical mobilization, z = −0.402 (Table 3, Figure 4).

µg/dL: microgram per deciliter, ms: milliseconds, n: population, p: adjusted significance, r: effect size, SD: standard deviation, * statistically significant p < 0.05

Discussion

There is an indication of a differential physiological response between upper and lower cervical spine mobilization at different timepoints, reflecting the complexity of interactions between the ANS and HPA-axis within the stress response. These results will be discussed below.

Salivary cortisol (primary outcome)

There was a statistically significant within-group decrease in sCOR from pre-5 to post-30 following lower cervical mobilization, and a statistically significant between-group difference in sCOR at the post-30 timepoint (Figure 3). This indicates a downregulation of cortisol secretion from the HPA-axis 30 min following lower cervical mobilization. In support of our findings, a thoracic manipulation performed on healthy men aged 18–45 reduced sCOR concentration at 5- and 30-min following the intervention [26]. The thoracic spine has a close anatomical relationship with the sympathetic nervous system (SNS) [91], as does the lower cervical spine with the stellate ganglion of the SNS [80]. Considering the bidirectional and interchangeable communication between the peripheral components of the stress response [7], the lower cervical mobilization and thoracic manipulation could have a similar effect on the HPA-axis.

Albeit non-significant, there is indication of a differential direction of response from pre-5 to post-5, where upper cervical mobilization increases sCOR, and lower cervical mobilization reduces sCOR. Despite basing the primary outcome on the average time it takes cortisol to enter saliva from blood [54], it can enter plasma much quicker via the fast corticosteroid feedback loop. The corticosteroid feedback loop operates in at least three domains; fast (seconds to minutes), intermediate (2 to 10 h), and slow (hours to days) domain [92]. Upper cervical mobilization may stimulate the superior cervical ganglion located at the level of the second and third cervical vertebrae, which is sympathetic [80], more so than the PSNS via the brain stem located at the level of the first cervical vertebrae [78,79]. It is possible that our interventions activated the fast corticosteroid feedback loop, resulting in an increase in sCOR following upper cervical mobilization, and a decrease following lower cervical mobilization.

Plausibly, the differential response from pre-5 to post-5 may be due to the individuals baseline cortisol concentrations, not the underlying anatomy. Our results showed that a higher baseline cortisol in the lower group lead to a downregulation of the HPA-axis and reduction in sCOR concentration, whereas a lower baseline in the upper group lead to an upregulation of the HPA-axis and increase in sCOR concentration. These results are mirrored in another study, which concluded that an upper cervical manual therapy technique increased blood pressure in those with hypotension, and decreased blood pressure in those with hypertension [93]. Cortisol is essential for the maintenance of normal blood pressure [94,95]. This differential response indicates that manual therapy, to restore homeostasis, will either upregulate or downregulate its response depending on the individual’s baseline. This highlights the function of the stress response, which works bidirectionally and interchangeably to maintain homeostasis through feedback loops [7].

There was a trend for sCOR concentrations to increase the night following each intervention (Figure 4). This increase was greater following lower cervical spine mobilization. The same trend was shown following a thoracic manipulation, where there was an initial decrease in salivary cortisol, followed by a reciprocal increase 6 h later [26]. Our results mirror this for the lower cervical mobilization group, where there was an initial decrease in sCOR concentrations, followed by a reciprocal increase that night. This response has been termed the ‘rebound effect’ [96]. Considering the increase in cortisol concentrations the night following the intervention, mobilizations applied to the cervical spine may be novel treatment for clinical populations characterized by low cortisol levels, such as persistent post-concussion symptoms [97–101] and long-COVID [102–104].

Heart rate variability (secondary outcome)

There was a statistically significant decrease in rMSSD the morning following upper cervical mobilization (Figure 4). Although non-significant, the same trend was shown following lower cervical mobilization. A decrease in rMSSD represents an attenuation of ANS regulation and reduced ability to adjust the internal environment in response to stressors [105,106]. Considering the interconnected relationship between the ANS and HPA-axis [7], and the trend for sCOR concentrations to increase in both groups on the night of the intervention, it may indicate that individuals are in a less parasympathetically dominant state due to PSNS withdrawal [106,107]. Our results indicate that this withdrawal was still present the morning following the intervention.

There was a trend for a differential response in rMSSD at post-5 and post-30, depending on the mobilizations target location (Figure 3). These findings are corroborated by the function of ANS within the context of the stress response; the ANS being the fast-acting component [1,4–6]. The ‘rebound effect’ concept is supported by the upper cervical group’s response, where there is an initial increase at post-5, followed by a reduction at post-30 which continues until the following morning. Other studies, albeit statistically significant, support our findings of upper cervical mobilization stimulating the PSNS [16,29], and lower cervical mobilization stimulating the SNS [29]. The statistically significant results found in these studies may be due to differing tools (ECG) and metrics (i.e. LF/HF, SDNN) used to measure HRV, different interventions (manipulation), or different populations (including females).

Strengths and limitations

A major strength of our study is its internal validity. A stringent inclusion criterion, and careful account/control of many variables shown to influence our outcomes of interest, help strengthen our results. This does, however, limit our study’s external validity. Baseline-pm sCOR concentrations fit within normative age matched values [62,108], and pre-5 concentrations fit within each individuals baseline-afternoon values and normative age matched values [62], further supporting accuracy of our results. Missing data were minimal for each analyzed participant (3.5% for sCOR, and 2.8% for rMSSD) and there was no attrition of participants. Another strength of this study is that it examined both components of the stress response, the ANS and HPA-axis, in unison.

This study is not without limitations. Our study was powered for the primary outcome, sCOR, to detect a large effect size, but it was only moderate. This may have introduced a type II error. Further to this, a type II error is possible due to our small sample size and large variability between participants evident by the 95% confidence intervals (Figures 3 and 4). The crossover design should reduce the requirement for a larger sample size by mitigating individual variability [109,110]. Regardless, future research should consider adequate sample sizes and account for the various factors which can modulate the ANS and HPA-axis (Appendix 4). Non-normally distributed data meant we could not perform statistical analyses to obtain a group x time interaction. Variability within our outcomes and small sample size likely accounts for the non-normal distribution. Considering the small sample size, it is possible that baseline (pre-5) sCOR variation accounted for the statistically significant results in clinic (Figure 3) [111]. However, statistical analyses revealed no statistically significant difference in pre-5 sCOR concentrations between groups, and the pre-5 values correlated with the population’s afternoon-pm baseline values, and age matched normative values [62].

The aim of this work was not to investigate efficacy, but to investigate whether there was a differential response, so there was no sham or control group. Specific to a crossover trial, and considering our small sample size, there is the possibility of a carryover, sequence, or period effect [111]. Statistical analyses of our statistically significant results ruled this out [89]. A non-stressed population may not detect the same magnitude or direction of response as it would in a stressed population, limiting the generalizability of our results.

Implications for clinical practice and future research

This study has provided a starting point for the rationale of using manual therapy more selectively to target specific mechanisms. Aligning differential responses to underlying pathophysiology opens the way to more efficient and effective treatment. For example, persistent post-concussion symptoms can be characterized by low cortisol, which has been associated with more symptoms, more severe symptoms, and delayed recovery [97–101]. Based on the present study results, we could perform lower cervical mobilization to increase cortisol levels that night, which could potentially lead to clinical benefit. A logical next-step from this study would be investigating the differential response of cervical spine manual therapy on clinical populations characterized by a dysfunctional stress response, such as those with persistent post-concussion symptoms [97–101], chronic pain [112] or osteoarthritis [113], and relating the physiological changes to clinical outcomes. Healthy patients and those with neck pain showed a differential HRV response to lower cervical manipulation [29]. Additionally, future studies could build on this trial by measuring our outcomes over a longer period of time and comparing the time-course effects of cervical spine mobilizations with other modalities shown to modulate the stress response, such as exercise [114], vagus nerve stimulation [115], or breathing control [116]. Cortisol is a pleiotropic hormone effecting all major homeostatic systems, such as the immune, cardiovascular, central nervous system, therefore its mechanisms of effect will likely be very widespread [31]. Future research needs to investigate the symptoms relating to a dysfunctional stress response, and responders/non-responders to a mechanisms-based intervention, termed ‘clinical phenotyping’ [117].

Conclusion

Our study has shown that manual therapy applied to separate target locations within the cervical spine can differentially modulate the stress response. Thirty minutes following lower cervical mobilization, there was a statistically significant within-group reduction in cortisol concentration. There was also a statistically significant difference in cortisol concentrations between groups at 30 min following upper and lower cervical mobilization. These findings highlight the intricate, time dependent, interaction between both components of the stress response and the complexity of investigating such phenomenon.

Author contribution

Concept development: GF, CCh, EK, KS, CCo, ST

Literature search: GF

Design: GF, MB, CCh, EK, KS, CCo, RK, ST

Recruitment: GF

Interventions and participant management: GF and ST

Data collection: GF and ST

Data analysis: MB, RK, GF

Manuscript preparation: GF

Commented/edited final manuscript: GF, MB, CCh, EK, KS, CCo, RK, ST

Disclosure statement

No potential conflict of interest was reported by the authors.

SUPPLEMENTARY MATERIAL

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10669817.2023.2177071

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The work was supported by the Stanley Paris PhD OMT Scholarship; and the Mark Steptoe Memorial Trust Fund

Notes on contributors

Gerard Farrell

Gerard Farrell is a PhD candidate at the Center for Health, Activity and Rehabilitation Research, School of Physiotherapy at the University of Otago. His research interests include the neuroendocrine mechanisms of manual therapy, the part a dysfunctional stress response has to play in the development and maintenance of persistent post-concussion symptoms, and the role manual therapy plays in the treatment of persistent post-concussion symptoms.

Matthew Reily-Bell

Matthew Bell is a PhD candidate at the Department of Physiology, HeartOtago, School of Biomedical Sciences. His PhD is investigating CRISPR/Cas13 as a tool for the knockdown of microRNA unregulated in diabetic heart disease.

Cathy Chapple

Dr. Cathy Chapple is a senior lecturer at the Center for Health, Activity and Rehabilitation Research, School of Physiotherapy at the University of Otago. Her research interests are in the management of people with osteoarthritis. She is also investigating manual therapy for osteoarthritis patients and other musculoskeletal conditions including the role of the cervical spine in persistent post-concussion symptoms.

Ewan Kennedy

Dr. Ewan Kennedy is a senior lecturer at the Center for Health, Activity and Rehabilitation Research, School of Physiotherapy at the University of Otago. His research interests are in musculoskeletal disorders, concussion, and clinical education. Recent work has explored overlap between cervical spine and concussion injuries, with a focus on improving health service delivery.

Kesava Sampath

Dr. Kesava Sampath is a senior lecturer at the Center of Health and Social Practice, Waikato Institute of Technology, Hamilton, Waikato, New Zealand. His research work focuses on mechanisms underpinning manual therapy practice. Currently, he leads a research project investigating the usage of bio-psycho-social model of care by New Zealand osteopaths while treating people with persistent musculoskeletal pain. He is a visiting research fellow at the University of Technology, Sydney, where he is part of a global osteopathic network and actively collaborates/contributes to other projects. He is also an adjunct researcher at the Ara Institute of Canterbury and supervises post-graduate nursing research students.

Angela Spontelli Gisselman

Dr. Angela Spontelli Gisselman is an assistant professor at Tufts University School of Medicine. Her main research interests are in the role of health metrics, such as heart rate variability (an index of the autonomic nervous system), and their ability to influence decision making in rehabilitation, such as post-concussion rehabilitation. In addition to this line of research, other areas of interest include heart rate variability and temporomandibular disorders; the use of thermal imaging in tendinopathy; examination and management of shoulder injuries; load monitoring technology and post-operative rehabilitation.

Chad Cook

Prof. Chad Cook is a clinical researcher, physical therapist, and profession advocate with a history of clinical care excellence and service. His passions include refining and improving the patient examination process and validating tools used in day-to-day physical therapist practice. Dr Cook has authored or co-authored three textbooks, has published over 250 peer-reviewed manuscripts, and lectures internationally on orthopedic examination and treatment. He is a fellow of the American Academy of Orthopaedic Manual Physical Therapists; has specialized in manual therapy for over 19 years. His main areas of research interest are in examination and conservative or surgical treatment of orthopedic-related conditions.

Rajesh Katare

Assoc. Prof. Rajesh Katare is an associate professor at the Department of Physiology, HeartOtago, School of Biomedical Sciences. His main research interests include investigating the molecular mechanisms involved in the development of cardiovascular complications in diabetes, the role of microRNAs in cardiovascular diseases, the development of novel genetic and stem cell therapies for the treatment of ischemic and non-ischemic cardiovascular complications.

Steve Tumilty

Assoc. Prof. Steve Tumilty. is one of the few clinician scientists in Physiotherapy Worldwide and one of the only nine Registered Physiotherapy Specialists in New Zealand. He has an interest in blurring the boundaries between the manual therapy professions. The majority of his clinical experience has been in the outpatient musculoskeletal practice setting in UK, Germany and New Zealand. He also has experience in professional sports and Occupational Health Physiotherapy. In 2002, he came to work at the School of Physiotherapy, Otago University and he has developed and now coordinates the specialist Masters degree for Sports and Orthopaedic Manipulative Therapy for which he provides teaching and clinical expertise. Dr Tumilty’s current research interests are in Tendinopathy, photobiomodulation, modulation of the Hypothalamus-Pituitary Axis using manual interventions, and the influence of the autonomic nervous system on musculoskeletal pain and healing.