Our study was approved by the RMIT Human Research Ethics Committee. Written, informed consent was obtained from participants before commencement of experiments, and all study protocols were conducted in accordance with the Declaration of Helsinki.
Eligible participants were between 18 and 35 years old and in good general health. Nineteen young adults responded to advertisements placed around the RMIT University campus, but four were excluded: due to high blood pressure (two), medication use (one), and benign arrhythmia (one). Participants (nine males and six females) were 24 ± 3 years old and had a body mass index of 22.2 ± 3.5 kg/m2 (expressed as mean ± standard deviation [SD]). None were smokers, used medication, or had a history of cardiovascular disease, diabetes mellitus, or cancer. Prior to the experiment, to assess general health status and account for factors that might influence autonomic and cardiovascular activity, participants completed general health, cardiovascular, and pre-experimental questionnaires. These focused on medical history, current health status, tobacco and medication use, and food and caffeine intake. Participants also completed questionnaires after each experimental session, regarding unpleasant sensations or discomforts during the experiment. Discomfort was assessed using a 10 cm, visual analogue scale (VAS), where 0 indicated "complete comfort" and 10 "worst pain imaginable."
Measurement of autonomic function
Heart rate (HR), HRV, and systolic and diastolic blood pressure (BP) were measured. A 3-lead electrocardiogram (ECG) allowed measurement of quick changes in HR , and visualisation of the QRS waveform. Disposable electrodes (Blue Sensor, Medicostest, Denmark) were positioned, with the negative electrode over the manubrium and the positive and earth electrodes at the left and right axillary lines (over the 5th intercostal space). Signals were amplified (BIO Amp ML 132, ADInstruments, Castle Hill, NSW, Australia) and stored on a personal computer. R-R intervals were calculated (Chart for Windows V 5.1.1 with HRV extension V 1.0.1, ADInstruments, Castle Hill, NSW, Australia) and the power spectrum of HRV was derived for the period of each intervention. The high frequency (HF) (0.15-0.4 Hz) component of the HRV power spectrum reflects parasympathetic activity  and the low frequency (LF) (0.04–0.15 Hz) reflects a combination of sympathetic and parasympathetic activity . The ratio of LF to HF (LF/HF) was adopted to determine the predominance of cardiac sympathetic nervous activity. We did not calculate the power of the very low frequency component (0–0.04 Hz), because it is unreliable over short recording periods .
Measurement of cardiovascular function
A Portapres® (Model-2, Finapres Medical Systems, The Netherlands) continuously measured BP and HR, using a finger cuff around the middle finger of the right hand. The Portapres® uses a hydrostatic height correction to transform measured BP values to those expected at the level of the heart (cf. ). Results were transferred to the data acquisition system (Chart for Windows) and displayed on a computer monitor, in real time.
Autonomic and cardiovascular functions were measured during prone, supine, and sitting postures. Participants were encouraged to position themselves comfortably, but once settled, were asked to remain still for recording.
Participants laid horizontally on a treatment table with hands on hand rests. The headrest was designed to facilitate participants' breathing and was adjusted to minimise neck flexion, extension, and rotation.
Participants lay on the table with a contoured pillow supporting their natural cervical lordosis.
A custom-designed chair supported participants' upright posture while minimising body and head movement. Footrests permitted comfortable knee flexion and both seat cushion and back support were provided. Immediately prior to recording, a helmet frame fixed the participant's head in a neutral position.
Recordings of HR and BP in the three postures were made in an air-conditioned laboratory, with white noise minimising disturbing sounds. Participants were asked to abstain from food and caffeine-containing beverages for at least 4 hours prior to data collection, and from alcoholic beverages and exercise for at least 12 hours. Two experiments (prone versus supine and prone versus sitting) were conducted on different days. To help minimise diurnal variation, participants were encouraged to schedule each experiment for the same time of day.
The vestibular system is responsible for balance , and is thought to influence autonomic and cardiovascular activities [22–24]. Therefore, to minimise vestibular organ activation, participants were instructed to avoid head motion during recording; they were also encouraged to stay awake. Adjustment of autonomic function to a particular posture is thought to occur within 5 minutes . Therefore, to stabilise autonomic outflow to cardiovascular organs before definitive recordings for each posture, participants were asked to make themselves more comfortable, and then remain still for 5 minutes. Additionally, through respiratory sinus arrhythmia, a participant's respiratory rhythm can influence HRV components . To standardise this impact, participants were asked (following the rest period) to synchronise their breathing to a metronome set at 0.25 Hz (15 times a minute) for 5 minutes.
Day 1: prone-supine
In the prone posture, HR and BP were measured continuously during both resting and breath-synchronised phases. Participants then moved to a supine posture, and recordings were repeated.
Day 2: prone-sitting
Identical to Day 1, except that prone posture was followed by sitting posture.
To confirm normal autonomic nervous function , at the end of Day 2, participants were asked to place a hand in a bucket of icy water (the cold pressor test), for as long as they could tolerate, but not longer than 1 minute. Blood pressure and HR were monitored during the test; although not required, had a significant sudden drop in BP and/or HR been observed, we would have terminated the procedure and excluded the participant from the study. Participants were also excluded if they did not respond to this test.
We recorded HR, mean arterial pressure, and systolic and diastolic BP, during rest and synchronised breathing periods in each posture. Mean values of each parameter were computed using Chart for Windows and analysed with the statistical software package SPSS (V 12.0.1 for Windows, SPSS Inc., U.S.A).
Electrocardiographic data recorded during synchronised breathing periods in each posture were analysed off-line (Chart V 5.1.1 with HRV extension for Windows V 1.0.1) for frequency spectrum characteristics, including LF and HF (absolute and normalised), and LF/HF. Paired samples t-tests were used to compare postures. When measurements for a variable deviated markedly from the normal distribution, the Wilcoxon signed-rank test was used (and reported as z scores). The statistical significance level for each comparison was set at p < 0.05.
Reproducibility of cardiovascular and autonomic parameters on different days was examined via the intraclass correlation coefficient (ICC). Values above 0.75 were considered to indicate good reliability, lower values poor to moderate . Paired samples t-tests were conducted to check for consistent differences in these parameters across recording days.
To evaluate cold pressor test response, minimum values of BP and HR were compared against mean values of BP and HR recorded during synchronised breathing periods in the sitting posture. A normal response to the cold pressor test was defined as a change in BP and/or HR to at least 2 standard deviations of reference values.