What is an 'optimal' spinal position during sleep? A systematic review

      Purpose: The mattress industry is fast out-growing the research behind it; the global market was recently valued at USD 27 billion. Mattress companies boast of innovative and stylish designs, assuring consumers of a good night’s sleep. Advertisements often include infographics highlighting a sleeper’s ‘optimal’ spinal curvature whilst lying on their product. Similarly, the research in this field hypothesises an optimal spinal position when sleeping: which is akin to that of standing. This is largely based on a theory that this posture minimises stress on spinal tissue and therefore improves sleep quality. A previous review concluded that medium-firm and custom-made mattresses improve sleep quality. Whilst this is important, focusing solely on the composition and dimensions of a mattress is only one half of the answer. This review focuses on the sleeper lying onthe bedding, or more specifically their spinal position. The study of the properties of a sleeping surface and the force it exerts on the spine crosses from the medical into the ergonomics / bioengineering fields. Previous reviews have not included engineering databases in their search strategies. The purpose of this systematic review is: 1. to examine the evidence underpinning the hypothesis that an optimal spinal position during sleep is akin to that of standing; and 2. to ascertain whether this posture correlates with improved sleep quality.
      Methods: Medical and engineering databases were searched including Medline, PEDro and IEEE Xplore. Studies from 1990 to 2018 were included, including populations with and without spinal pain or deformity. Studies which measured spinal position in lying were included. Studies which did not specifically measure this, but involved a mattress design which presumed an optimal spinal position of the subject, were also included. A snowballing technique was performed within each article’s reference list to interrogate the evidence behind the hypothesised optimal spinal position.
      Results: Five studies examined optimal spinal position (table 1). Of them three studies measured stress on spinal tissue only in an erect spine, not a supine position. The rest (two studies) measured tissue stress in a lying position. No comparison was made between optimal and sub-optimal spinal posture. All five studies were published prior to the mid 1990s, and all were of low quality. Eight studies measured sleep quality in different spinal positions (table 2). One study included subjects with spinal pain and/or deformity; the remainder of subjects were healthy and pain-free. Four studies simply presumed an optimal spinal position of the subject based on the mattress design, whilst only four explicitly measured spinal position. The heterogeneity of spinal position measurement techniques, as well as a lack of supporting validation studies, renders it difficult to directly compare results. The duration of the intervention varied significantly between studies, ranging from 30 seconds to nightly for two weeks. The lying position adopted by subjects also varied. In four studies of shorter experimental duration (30 seconds to 30 minutes) subjects were required to adopt one single position (i.e. either supine or lateral-lie). Four studies more closely mimicked physiological sleep, measuring subjects for at least one night without placing restrictions on sleeping position. Six studies found improvements in at least some measurements of sleep quality with an optimal spinal position; five of which found no change in terms of other measurements. Two studies demonstrated no change in any measurements of sleep quality. One study further specified that subjects with spinal pain or deformity had significantly improved sleep quality when adopting an optimal spinal posture, compared to those without. The heterogeneity of measurements of sleep quality between studies, however, renders it difficult to directly compare results.
      Conclusions: The optimal position for good quality sleep is unclear. There is minimal, low-quality data examining whether this position is akin to that of standing. Given the importance of good quality sleep, and a growing mattress industry, further work is needed to guide mattress design and inform consumers as to whether an optimal spinal position exists - and what that is. More high-quality studies with consistent, validated measurement tools are required to answer this question. Then, further work including populations with back pain or spinal deformity would be useful, as this group may benefit from attention to spinal position in sleep.
      Nachemson et al (1964)In vivo experimental studyStanding, unsupported sitting, reclining positionsIntra-discal pressure is highest in unsupported sitting (ie. in some degree of lumbar flexion), approximately 30% less in standing (ie. in lumbar lordosis) and approximately 50% less in reclining
      Nachemson (1965)In vivo experimental studyStanding and sitting positionsIncreasing lumbar flexion by 20 degrees increases intra-discal pressure by 30%. In 20 degrees lumbar flexion the L3/4 disc carries a load of 180-230kg. In this position when carrying 10kg the load in the L3/4 disc is 250-340kg.
      Wilke et al (1995)In vivo experimental studyVarious lying, sitting, standing and erect dynamic posturesL4/5 disc sustained: 0.1MPa pressure when lying prone; 0.12MPa pressure when lying laterally; 0.3MPa pressure in nonchalant sitting; 0.46MPa pressure when sitting unsupported; 0.5MPa pressure in relaxed standing; 0.83MPa pressure when sitting in maximum flexion; 1.1MPa pressure when standing flexed forward; 1.1MPa pressure when lifting a 20kg weight close to the body; 1.7MPa pressure when lifting a 20kg weight with flexed knees; 2.3MPa pressure when lifting a 20kg weight with a round flexed back. During the night L4/5 pressure increased from 0.1 to 0.24MPa
      Adams et al (1980)Cadaveric experimental studyErect lumbar lordosis (to mimic a standing position) and erect lumbar kyphosis (to mimic a sitting position)In erect lumbar lordosis the zygapophyseal joints absorb 16% of axial load. In erect lumbar kyphosis the zygapophyseal joints absorb 0% of axial load.
      Bennet et al (1989)Controlled trialStanding position and three different chairs promoting different amounts of lumbar lordosisEMG activity of paraspinal musculature in standing is greater than in sitting in three different chairs. No difference in EMG activity across any of the chairs. No correlation between EMG activity in paraspinal musculature and lumbar curvature.
      De Vocht(2006)Four generic mattressesPostural distortion (photography and imaging digitizing software): Mattress Dcaused significantly lower overall spinal distortion (mean slope = 0.103 ±0.106) compared to B, A and C(0.115 ±0.098, 0.118 ±0.102 and 0.120 ±0.108, respectively).Maximum pressure (mmHg): Mattress Dcaused significantly greater maximum pressure (38.224±5.687) compared to A, B, & C (29.2±5.6, 30.546±5.776 & 32.561±4.832)
      Derman (1995)Generic lumbar body support atop a conventional mattress vs Conventional mattress aloneSpinal position not measured. The use of a generic lumbar body support in the supine position is presumed to promote a spinal position akin to standing.Heart rate; subjects with pain; acute phase: 69±1 vs 75±5 bpm, p<0.05. Heart rate; subjects with pain; chronic exposure: 69±1 vs 75±5 bpm, p<0.05. Discomfort ratings; subjects with pain; acute phase (scale 0-10): 2±0.4 vs 8±0.5, p<0.05. Discomfort ratings; no pain; acute phase (scale 0-10): 1±0.1 vs 7±0.7, p<0.05. Discomfort ratings; subjects with pain; chronic exposure (scale 0-10): 1±0.1 vs 7±0.7, p<0.05. Discomfort ratings; no pain; chronic exposure (scale 0-10): 2±0.2 vs 3±0.2, p<0.05. EMG activity; subjects with pain; acute phase: 5.5 vs 8.5±1.5 microvolts, p<0.05. EMG activity; subjects with pain; chronic exposure: 5 vs 8±2 microvolts, p<0.05.Heart rate; no pain; acute phase: 69±1 vs 62±7 bpm, no significant difference (value not listed). Heart rate; no pain; chronic exposure: 61±6 vs 60±7 bpm, no significant difference (value not listed) . EMG activity; no pain; acute phase: 5.5 vs 5 microvolts, no significant difference (value not listed) . EMG activity; no pain, chronic exposure: 5.5 vs 5.25 microvolts, no significant difference (value not listed).
      Lahm & Iazzo(2002)Generic air inflatable mattress of three increasing pressures(827.4Pa, 2413.2Pa and 3999.0Pa)Spinal displacement from horizontal axis (using digital photography): 8871.0 ±3632.3 mm2, 6632.3 ±3109.7 mm2and 5703.2 ±3180.6mm2respectively, p < 0.01.Spinal displacement from a straight line drawn between end points (using digital photography): 2903.2 ±1509.7, 2290.3 ±1206.5 and 2116.1 ±1135.5mm2, p = 0.1125.Heart rate: 68.2 ±9.5, 67.8 ±9.6 and 69.0 ±10.1 beats/min. Blood pressure: 96.4±10.6, 94.6±10.3 and 94.4 ±7.9 mmHg. Subjective discomfort ratings (10 point scale):6.8±1.8, 7.1±1.3 and 6.4±1.6. EMG activity (aggregate % change): 28.7±9.9%, 28.4±9.8% and 25.5±8.2%. No significant difference (value not listed). Mattress Interface Pressure Profile: 13.0±0.9 mmHg, 13.0 ±1.1mmHg and 13.0 ±1.1 mmHg. No significant difference (value not listed).
      Normand et al (2005)Generic lumbar support in three conditions (no mattress, under 8cm foam, under 14cm latex mattress) vs Above conditions without lumbar supportSpinal position not measured: The use of a generic lumbar body support in the supine position is presumed to promote a spinal position akin to standing.Mean contact pressure in the lumbar area: No mattress:174.19±25.24 vs 11.92±4.95 N; Foam:41.62±8.64 vs 26.76±5.85 N; Latex: 164.3±17.3 vs 53.63±12.71 N; (F1,9=70.021, p<0.001). Mean contact pressure in the pelvic area: No mattress:388.48±30.95 vs 448.86±31.56 N; Foam:71.71±10.11 vs 79.71±11.72 N; Latex:94.05±17.61 vs 115.7±20.86 N; (F1,9 = 47.967, p<0.001). Mean contact pressure in the thoracic area: No mattress:60.8+26.11 vs 106.92±35.58; (F2,18) = 11.916 p < 0.001.Mean contact pressure in the thoracic area: Foam:43.33±10.02 vs 43.81±10.52 N; Latex:117.24±26.46 vs 107.77±27.78 N; No significant difference (value not listed).
      Park et al (2001)Six generic mattresses of increasing firmnessSpinal position (plaster cast and 3D measurement tool): Results displayed graphically.Subjective comfort rating: ‘Comfortable’ mattresses accommodated spinal curvatures of lower root mean square of 28.53±12.97. ‘Uncomfortable’ mattresses accommodated spinal curvatures of higher root mean square 55.0+29.54.
      Van Deun et al (2012)Custom mattress of eight automatically, dynamically adjusting zones vs Same mattress without automatic featureSpinal position not measured: Custom mattress presumed to promote a spinal position akin to standing. This mattress was designed to measure indentation in eight different zones and therefore sleep posture. Comparison was made between estimated and desired spinal shape, and the eight zones were automatically adjusted by actuators applying vertical displacement of the zones’ spring bases to achieve desired spinal shape. This control loop was repeated continuously throughout the testing night.Subjective sleep quality (scale 0-10): 7.5 ±1.32 vs 6.6 ±1.63, p = 0.029; Number of awakenings:0.80 ±0.79 vs 1.70 ±0.94, p = 0.041; Duration of awakenings:0.90 ±0.88 vs 1.25 ±0.63, p = 0.045; Daytime quality score (scale 0-30): 22.35±4.86 vs 24.36 ±4.4, p = 0.039; Time spent in N1 sleep (%):9.86 ±5.77% vs 7.91 ±5.12%, p = 0.027; Time spent in slow wave sleep (%): 29.89 ±6.19 vs 24.50 ±5.89, p = 0.096.Profile of mood state fatigue difference score (scale -5-5): ± 0.76 vs 0.15 ± 0.67, p=0.623; Activation/deactivation adjective checklist difference score (scale -5-5): -0.53 ± 1.08 vs -0.34 ± 1.04, p=0.554. Karolinska sleepiness scale difference: evening before vs morning after (scale -5-5): -0.95 ± 1.01vs -0.72 ± 1.07, p=0.513; Bedding system quality (scale 0-10): 4.95 ± 1.68 vs 4.53 ± 1.65, p=0.310; Feeling rested score (scale 0-10): 7.24 ± 1.92 vs 6.71 ± 1.87, p=0.248; Back pain score (scale 1-4): 2.39 ± 1.40 vs 1.86 ± 1.56, p=0.180; Sleep onset latency: 80 ±0.79 vs 1.10 ±0.88, p=0.430; Total sleep time (min): 423.55 ± 30.9 vs 423.5 ± 30.90, p= 0.68; Intra-sleep awake time (min): 34.29 ± 30.16 vs 37.37 ± 30.16, p= 0.44; Time awake (%): 4.44 ± 5.99 vs 8.07 ± 5.22, p= 0.56.
      Verheart et al(2011)Custom mattress of eight zones vs Custom mattress built to mimic a sagging mattressSpinal position not measured: Custom mattress presumed to promote spinal position akin to standing. This mattress was designed by with eight adjustable zones of varying stiffness to minimise deformation of the spine in a lateral posture (i.e. when the spine approximates a straight line in the frontal plane).Subjective sleep quality: overall (scale 0-20): 12.1±3.8 vs 14.4±1.8, x2(1) = 4.57, p<0.05; Subjective sleep quality: lateral/prone sleepers (scale 0-20): 9.0±3.1 vs 13.4±1.8, x2(1)=7, p<0.01; Cox’s Stress Arousal Adjective Checklist: lateral/prone sleepers: -0.6±0.9 vs. 1.0±0.7; x2(1)=7, p<0.01; Time spent awake (% of time in bed): lateral/prone sleepers: 12.1±5.1% vs 8.8±5.8%, x2(1)=6, p < 0.05; Awakening durations: lateral/prone sleepers: 4.0±1.3 min vs 2.4±1.2 min, x2(1) = 6, p < 0.05; REM sleep (%): lateral/prone sleepers: 11.5±6.2% vs 17.1±6.4%, x2(1)=6, p<0.05; REM latency: lateral/prone sleepers: 150.8±62.1 min vs 101.4±34.5 min, x2(1)=6, p<0.05.Subjective sleep quality: lateral/supine sleepers (scale 0-20): 14.9±1.4 vs 15.2±1.4, x2(1)=0.14, p=0.71; Cox’s Stress Arousal Adjective Checklist: lateral/supine sleepers: 0.6±1.1 vs 0.8±0.8; x2(1)=0, p=0.99; Sleep onset latency: lateral/prone sleepers: 24.75±11.39 vs 20.08±9.40 min, p=0.99; Total sleep time: lateral/prone sleepers: 374.33±18.39 vs 408.5±25.57mins, p=0.1; Number of awakenings: lateral/prone sleepers: 21.17±7.52 vs 25.50±4.85, p=0.1; Number of awakenings: lateral/supine sleepers: 20.25±6.76 vs 21.62±6.84, p=0.26; Awakening durations: lateral/supine sleepers: 1.96±0.86 vs 1.84±1.20, p=0.48; Sleep Efficiency Index (total sleep time/time in bed, %): lateral/prone sleepers: 84.2 + 5.50 vs 86.97 + 5.95, p=0.1; Sleep Efficiency Index (total sleep time/time in bed, %): lateral/supine sleepers: 92.42 + 4.94 vs 93.21 + 4.33, p=0.48.
      Verhaert et al(2013)Custom mattress of eight automatically, dynamically adjusting zones vs Same mattress without automatic featureSpinal displacement from neutral measured by the mattress itself (represented by five angles and two distances): Left lateral position: P1 1.07±1.22ovs 5.83±0.91o, P2 1.10±1.18mm vs 3.45±1.87mm, P3 1.25±1.25ovs 6.74+1.23o, P4 0.83±2.32ovs 6.01±4.00o, p<0.01;Right lateral position:P1 1.21±1.63ovs 6.52±1.57o, P2 0.94±1.10mm vs 3.82±2.86mm, P3 1.35±1.62ovs 7.46±1.82o, P4 1.29±2.20ovs 6.84±3.61o, p<0.01; Supine position:P5 6.54±2.35mm vs 8.59±3.51mm, p=0.03; P6 5.06±4.95ovs 6.67±5.87o, p=0.41; P7 2.80±2.01ovs 4.45±2.99o, p=0.24.Sleep quality (10 point scale):7.00±0.87 vs 5.67±1.41, p=0.02.Refreshed feeling (10 point scale):7.00±1.41 vs 6.11±1.83, p=0.06.