Posture and belt fit in reclined passenger seats

Abstract

Objective: Highly reclined postures may be common among passengers in future automated vehicles. A laboratory study was conducted to address the need for posture and belt fit in these seating configurations.
Methods: In a laboratory vehicle mockup, the postures of 24 men and women with a wide range of body size were measured in a typical front vehicle seat at seat back angles of 23°, 33°, 43°, and 53°. Data were gathered with and without a sitter-adjusted headrest. Posture was characterized by the locations of skeletal joint centers estimated from digitized surface landmarks.
Results: Regression analysis demonstrated that the pelvis rotated rearward and lumbar spine flexion decreased with increasing recline. The lap portion of the 3-point belt was more rearward relative to the pelvis in more-reclined postures, and the torso portion crossed the clavicle closer to the midline of the body. Regression equations were developed to predict posture and belt fit variables as a function of passenger characteristics, seat back angle, and the use of the headrest.
Conclusions: Spine posture changes as the torso reclines in an automotive seat, and belt fit is altered by the change in posture. The results can be used to accurately position crash test dummies and computation human models and to guide the design of belt restraints.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=gcpi20
Traffic Injury Prevention
ISSN: 1538-9588 (Print) 1538-957X (Online) Journal homepage: https://www.tandfonline.com/loi/gcpi20
Posture and belt fit in reclined passenger seats
Matthew P. Reed, Sheila M. Ebert & Monica L. H. Jones
To cite this article: Matthew P. Reed, Sheila M. Ebert & Monica L. H. Jones (2019) Posture
and belt fit in reclined passenger seats, Traffic Injury Prevention, 20:sup1, S38-S42, DOI:
10.1080/15389588.2019.1630733
To link to this article: https://doi.org/10.1080/15389588.2019.1630733
© 2019 The Author(s). Published with
license by Taylor & Francis Group, LLC
View supplementary material
Published online: 05 Aug 2019.
Submit your article to this journal
View Crossmark data

Posture and belt fit in reclined passenger seats
Matthew P. Reed, Sheila M. Ebert, and Monica L. H. Jones
University of Michigan Transportation Research Institute, Ann Arbor, Michigan
ABSTRACT
Objective: Highly reclined postures may be common among passengers in future automated
vehicles. A laboratory study was conducted to address the need for posture and belt fit in these
seating configurations.
Methods: In a laboratory vehicle mockup, the postures of 24 men and women with a wide range
of body size were measured in a typical front vehicle seat at seat back angles of 23,33
,43
,
and 53. Data were gathered with and without a sitter-adjusted headrest. Posture was character-
ized by the locations of skeletal joint centers estimated from digitized surface landmarks.
Results: Regression analysis demonstrated that the pelvis rotated rearward and lumbar spine flex-
ion decreased with increasing recline. The lap portion of the 3-point belt was more rearward rela-
tive to the pelvis in more-reclined postures, and the torso portion crossed the clavicle closer to
the midline of the body. Regression equations were developed to predict posture and belt fit vari-
ables as a function of passenger characteristics, seat back angle, and the use of the headrest.
Conclusions: Spine posture changes as the torso reclines in an automotive seat, and belt fit is
altered by the change in posture. The results can be used to accurately position crash test dum-
mies and computation human models and to guide the design of belt restraints.
ARTICLE HISTORY
Received 4 November 2018
Accepted 7 June 2019
KEYWORDS
Passenger posture; seat belt
fit; reclined seat
Introduction
Increasing road vehicle automation is expected to lead to
changes in passenger activities. A common prediction is that
highly reclined postures will become more common as passen-
gers choose to rest or sleep during travel. Currently, little is
known about the details of reclined passenger postures, includ-
ing typical body segment angles. Data-based posture prediction
models have been developed for passengers (Park et al. 2016a),
but these are limited to seat back angles (SAE A40) of 30.
Crash safety is a prominent concern with reclined postures.
Recent simulation studies have suggested that protecting passen-
gers in highly reclined postures is challenging due to differences
in occupant kinematics compared with more upright seating
(Lin et al. 2018), but the postures used in those studies were not
based on actual passenger data. Data on occupants in these seat-
ing configurations are needed so that crash simulations can be
performed using accurate representations of occupant posture
and belt fit. Both physical testing with anthropomorphic test
devices and computational human body modeling will benefit
from quantitative information from passengers.
To address this gap, the postures of a diverse group of
passengers were measured across a range of seat back angles
in a laboratory study. Data were gathered with and without
head support at seat back angles up to 53. Lap and shoulder
belt fit relative to the pelvis and clavicle were also quantified.
Methods
Twenty-four adults (12 men and 12 women) participated in
this study. The participants were chosen to span a wide range
of body size so that the potential effects of body size on pos-
ture and belt fit could be investigated. The mean stature of the
group was 1,694 mm with a range of 1,521 to 1,898 mm, mean
body mass index (BMI) was 28.9 kg/m
2
with a range of 21.0 to
37.2 kg/m
2
, and mean age was 47 years with a range of 18 to
71 years. Appendix Table A1 (see online supplement) presents
additional anthropometric data. Standard anthropometric
dimensions, including stature, body weight, and linear
breadths and depths, were gathered from each participant to
characterize the overall body size and shape. All data collec-
tion, including the measurement of standard anthropometric
dimensions, was conducted with the participants clad in light
cotton pants and shirt.
A vehicle mockup used in previous studies of driver posture
and belt fit (Reed et al. 2013; Park et al. 2016b) was modified for
use in the current testing. The mockup (Figure 1) was equipped
with a 6-way power seat from a 2010 Toyota Highlander with a
power recline adjuster. The seat pan angle was locked at 14.5as
measured using the SAE J826 manikin and procedures. The seat
was moved rearward to a position where the participant’sfeet
could not contact the pedals. The seat H-point was measured
using the SAE J826 manikin (SAE International 2018a)andthe
CONTACT Matthew P. Reed mreed@umich.edu University of Michigan Transportation Research Institute, 2901 Baxter Road, Ann Arbor, MI 48109.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcpi.
Managing Editor David Viano oversaw the review of this article.
Supplemental data for this article can be accessed on the publisher’s website.
ß2019 The Authors. Published with license by Taylor & Francis Group, LLC
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
TRAFFIC INJURY PREVENTION
2019, VOL. 20, NO. S1, S38–S42
https://doi.org/10.1080/15389588.2019.1630733

seatwasadjustedtoplacetheH-point270mmabovethe
mockup floor. The original head restraint was removed from the
seat. A “headrest base”was built from a padded board attached
to metal rods that inserted into the head restraint receptacles in
the seat back. The front surface of the headrest base was set back
from the seat back surface farther than a normal vehicle head
restraint to provide flexibility during testing. Note that this sup-
port is termed a “headrest”rather than “head restraint”because
it was not designed to be appropriate for rear impact protection
or to comply with head restraint requirements. Instead, the goal
was to determine appropriate geometry for comfortable head/
neck support for passengers.
A seat belt assembly with a sliding latch plate and retractor
from the second row of a model year 2010 Toyota Sienna was
mounted on customized fixtures designed to permit adjustment
of belt anchorage locations. A second-row belt was used to
ensure sufficient webbing length for all package conditions. A
rigid buckle stalk was attached to the seat with an adjustable fix-
ture. The outboard lower anchorage was attached to the
mockup, rather than to the seat, simulating a belt mounted to
the vehicle body. The retractor and D-ring were mounted to a
fixtureallowingtheD-ringlocationtobeadjustedoverawide
range. The belt webbing width was 45 mm.
The seat back angle (SAE A40, also known as manikin
torso angle) was initially set to 23as measured with the SAE
J826 H-point machine and procedures (SAE International
2018b). The seat back was rotated relative to this measure-
ment position to achieve a range of recline angles. The pivot
point of the seat back was 164 mm rearward and 86mm down
from seat H-point. The seat back was rotated 10,20
, and
30rearward to achieve nominal seat back angles of 33,43
,
and 53. Note that SAE A40 was only measured at 23.
The pivot point of the upper belt anchorage (D-ring bolt)
was located 312 mm rearward, 235 mm outboard, and
626 mm up from the seat H-point of the 23back angle
condition. The upper anchorage location was rotated around
the seat pivot point to keep it at the same location relative
to the seat back for each of the more reclined seat back con-
ditions. The seat pan and the lower belt anchorages were in
fixed positions. The lower anchorages were mounted at an
angle of 52up from horizontal with respect to the H-point
in the 23seat back angle condition.
The study protocol was approved by an institutional review
board for human subjects research at the University of Michigan
(HUM00142287). After giving written informed consent, the
participants changed into the testing clothes, underwent stand-
ard anthropometry testing, and then had their posture measured
in the test seat as well as on a laboratory hard seat.
Eighttestconditions(4seatbackangleswithandwithoutthe
headrest) were presented to the subjects in random order. A head-
rest was provided in half of the conditions. For each condition, the
investigator adjusted the seat back angle to 23and asked the par-
ticipant to be seated and to make him- or herself comfortable. The
investigator then reclined the seat back to the test condition angle.
The participant readjusted his or her posture if necessary for com-
fort and donned the seat belt. If the condition did not include the
headrest, the participants were asked to hold their head in a pos-
ition that allowed them to look straight forward. These postures
might be typical of a person looking out the windshield or viewing
an entertainment display. If the headrest was included, the partici-
pants placed one or more soft foam pieces of 23-mm-thickness
behind their heads until they achieved a posture that would be
comfortable for “a long rest with eyes closed.”
Posture was recorded by digitizing the location of body
landmarks using a FARO Arm coordinate digitizer (FARO
Technologies, Lake Mary, FL) and procedures identical to
those used in several previous studies (Reed et al. 2013; Park
et al. 2016a,2016b). In addition, a stream of points with
approximately 5-mm spacing was recorded along the edges
of lap and shoulder portions of the belt between the ancho-
rages and latch plate to facilitate quantification of belt fit.
Body landmark locations were also recorded in a labora-
tory hard seat that allows access to posterior spine and pelvis
landmarks that are inaccessible in the automotive seat (see
Reed et al. [2013] and Park et al. [2016b]) for more details on
the use of the hard seat). The hard seat has a 14.5“cushion”
(pan) angle and a 23back angle designed to produce postures
similar to those in an automotive seat. To estimate the pelvis
location, the adjustment for adiposity described in Reed et al.
(2013) was applied to the points recorded on the pelvis.
The position and orientation of the bony pelvis is difficult
to measure but of considerable importance. The methods
used in the current study are based on those reported by Park
et al. (2015). The methods rely on palpation and measurement
of the location of surface landmarks over the anterior superior
iliac spine (ASIS) and posterior superior iliac spine (PSIS)
landmarks on the pelvis. In brief, the steps are as follows:
1. Using the hard seat ASIS and PSIS locations, estimate
the subject-specific pelvis geometry using an estimate
Figure 1. Test seat with power recline showing user-adjustable headrest with
the seat back angle set to the 53condition.
TRAFFIC INJURY PREVENTION S39

flesh margins at the ASIS and PSIS that are based on
the adjustment for BMI presented by Reed et al. (2013).
The pelvis geometry is defined by surface ASIS, surface
PSIS, bone ASIS estimate, bone PSIS estimate, and esti-
mated L5/S1 and left and right hip joint centers. Also
record a “thigh length”as the distance between the
suprapatellar landmark and estimated hip joint center
location on each side obtained in the hard seat. Also
record the lumbar link length (distance from estimated
T12/L1 to estimated L5/S1 joint centers).
2. In the vehicle seat data, align the subject-specific pelvis
geometry to the measured mid-ASIS point and align the
lateral (inter-ASIS) axis.
3. Rotate around the lateral axis such that the lumbar link
length (distance from estimated the L5/S1 to T12/L1
joint locations) matches the hard seat value.
4. With this as a starting point, apply the optimization method
described in Park et al. (2015), which finds the pelvis pos-
ition and rotation around the lateral axis that best matches
both the lumbar and left thigh segment lengths. The con-
straints were adjusted from those used by Park et al. (2015)
to account for the reclined postures. Lumbar link length
waspermittedtodeviatefromthehardseatvaluebyupto
20 mm and the pelvis angle could change up to 10forward
or 45rearward from the initial value. Xand Ztranslation
constraints were 25 mm forward/45 mm rearward and
25 mm upward/50 mm downward. In all but 4 cases, the left
thigh segment length from the hard seat was matched
within 1 mm; in the remaining cases, the resulting thigh seg-
ment length was within 10 mm of the hard seat value.
Landmark data from the hard seat and vehicle seat were used
to characterize participant posture. Figure 2 illustrates the pri-
mary variables, which are based on the posture models reported
by Reed et al. (2002) and Park et al. (2016a,2016b). The torso
posture was defined based on a kinematic linkage consisting of
pelvis, lumbar, thorax, neck, and head segments. The pelvis seg-
ment connects the hip joints with L5/S1. The lumbar segment
spans L5/S1 to T12/L1 and the thorax segment connects T12/L1
with C7/T1. Note that these spine “joints”are defined as the
estimated center of the intervertebral disk (see Reed et al. 2002).
The neck segment spans C7/T1 to the estimated atlanto-occipi-
tal joint location. The mean hip location (average of left and
right hip joint centers) was computed with respect to the seat
H-point location. The side-view orientations of the pelvis,
lumbar, thorax, and neck segments were computed with respect
to vertical (positive values indicate reclined from vertical). Head
orientation was defined as the angle of the Frankfurt plane with
respect to forward horizontal (positive with eyes up). The thigh
angle was computed in side view with respect to forward hori-
zontal, and the leg angle was reported positive rearward of verti-
cal. Overall torso recline was quantified by the angle of the side-
view vector from mean hip to eye (estimate of the center of the
left eyeball) with respect to vertical.
Figure 3 shows the dimensions used to quantify belt fit.
The position of the upper/rearward edge of the lap portion
of the belt was recorded relative to the estimated ASIS land-
mark on the bony pelvis at the lateral position of the ASIS
(see Reed et al. [2013] and Park et al. [2016a] for additional
information on the calculation procedures). The position of
the torso portion of the belt was quantified by the lateral dis-
tance of the inner edge of the belt from the suprasternale
landmark on the body midline at the height of the landmark.
The effects of test conditions and participant characteristics
were examined using a range of plotting methods as well as lin-
ear regression and analysis of variance. Potential 2-way interac-
tions between participant characteristics and test conditions
were considered, and trends were examined within sex as well
as with the pooled male/female population. Potential sex differ-
ences were examined by testing for 2-way interactions between
sex and anthropometric variables such as stature and by restrict-
ing the analyses to a subpopulation of men and women with the
same range of stature. However, no sex-related effects were
found to be significant after accounting for body size. When
developing posture prediction models using regression, terms
were included only if they were significant with P<.01 and the
addition of the term increased the adjusted R
2
value (i.e., frac-
tion of variance accounted for by the model) by at least 0.02.
Figure 2. Landmarks, joints, and posture variables. Angles are positive as
shown except for neck angle.
Figure 3. Dependent measures for lap belt fit. The upper/rearward edge of the
lap portion of the belt is measured at the lateral position of the right and left
the predicted ASIS location. The fore–aft (X) coordinate is positive rearward of
the ASIS and the vertical coordinate is positive above the ASIS landmark.
S40 M. P. REED ET AL.

Results
Posture
The mean hip location with respect to seat H-point was not
strongly affected by either the experimental variables or partici-
pant characteristics. On average, the mean hip location was
11 mm below and 16 mm rearward of the seat H-point. The
mean hip location was significantly further rearward at the high-
est recline angle, but the difference relative to the other condi-
tions was less than half of the standard deviation within
condition. Appendix Table A2 (see online supplement) lists the
mean and standard deviation of hip location for each condition.
Tables A3 and A4 (see online supplement) list body seg-
ment angle results for the headrest and no-headrest conditions.
These tables summarize the variable values across subjects, so
the effects of body dimensions are not considered. Table A4
lists regression functions that incorporate anthropometric pre-
dictors. Results for several variables are summarized here.
Equation (1) shows the regression model for the overall
torso recline, represented by the side-view angle of the vec-
tor from the mid-hip location to the eye. Torso recline was
strongly affected by seat back angle (BA, ), as expected, but
also by BMI (kg/m
2
), with more upright angles observed
with higher BMI. On average, the use of the headrest (HR,
1j0) increased the value of this variable by 5.3.
HipEye Angle deg
ðÞ
¼7:5–0:217 BMI
þ5:3HRþ0:86 BA;RMSE ¼3:8p;R2adj ¼0:87:
(1)
Pelvis angle is an important variable for crash safety due to
the importance of pelvis engagement by the lap belt. Pelvis
angle was significantly affected by BMI, with slightly more
upright angles associated with higher BMI, but the residual
standard deviation of 15waslargerelativetothiseffect.As
expected, the pelvis tended to rock backward with increasing
back angle (larger pelvis angle values), but the change in pelvis
angle was only one-third the change in the seat back angle.
Pelvis Angle ðdegÞ¼ 84:8–1:37 BMI
þ0:331 BA;RMSE ¼15:4;R2adj¼0:19:
(2)
A measure of lumbar spine flexion was obtained by sub-
tracting the thorax angle from the pelvis angle. About 24%
of the variance in this measure was accounted for by the
participant covariates and experimental variables. On aver-
age, taller participants had slightly greater lumbar spine flex-
ion, and higher BMI was associated with lower flexion.
Using the headrest caused the thorax to rotate rearward,
decreasing lumbar flexion by 6.6. Increasing recline
decreased lumbar spine flexion substantially, because the
thorax rotated rearward more than the pelvis. On average,
an increase in the seat back angle of 30produced a 17
reduction in lumbar spine flexion.
Lumbar Spine Flexion ðdegÞ¼ 24:3þ
0:033 S –0:643 BMI –6:6HR–
0:591 BA;RMSE ¼15:2;R2adj¼0:24:
(3)
Figure 4 illustrates the effects of back angle on the mean
torso posture for conditions with the headrest, using the
mean anthropometric values (stature ¼1,694 mm, body
weight ¼84 kg). In the figure, the torso joint locations are
calculated using the angle regression functions in Table A5
(see online supplement). The increase in lumbar spine flex-
ion with increasing seat back angle is visible as the larger
change in thorax angle compared to pelvis angle.
Belt fit
Figure 5 shows the mean lap belt position relative to the
bone ASIS on the right side of the pelvis (see Appendix
Table A6, online supplement, for summary statistics). On
average, the lap belt was forward and above the pelvis. The
belt was further rearward relative to the pelvis, on average,
with increasing seat back angle, but the vertical position was
not significantly affected. Appendix Table A7 (see online
supplement) lists regression models for lap belt fit. BMI was
the dominant predictor, with higher BMI associated with
further forward and higher lap belt positions.
Figure 6 shows the distributions of torso belt score across seat
back angles. On average, the torso portion of the 3-point belt
was further inward with greater recline. The belt crossed the
body midline at the height of the clavicle at the largest recline
angle. Torso belt score was significantly affected by stature and
the ratio of sitting height to stature. A significant quadratic effect
of seat back angle was observed (see Appendix Table A7).
Discussion
This is the first detailed examination of highly reclined pos-
tures in automotive seats. Previous studies had been limited
to seat back angles of 30(Park et al. 2016a), but the current
data extend to 53and could reasonably be extrapolated to
60, based on the finding of predominantly linear effects of
seat back angle on posture variables.
This study is unusual in finding hip locations slightly
rearward of seat H-point, on average, even at the 23seat
back angle. Many previous studies have found hip locations
somewhat forward of the H-point (e.g., Reed et al. 2002;
Park et al. 2016a,2016b). More-rearward hip locations were
expected in the more-reclined conditions, because the pos-
ition of the seat back angle pivot results in the seat back
moving rearward behind the pelvis as it reclines. Because
Figure 4. Illustration of mean torso posture with headrest use across seat back angles.
TRAFFIC INJURY PREVENTION S41

the order of test conditions was randomized, the finding of
the hips being rearward of the H-point in the 23condition
may result from changes in participant behavior due to the
experience of highly reclined conditions.
The pelvis rotated rearward with increasing recline but
lagged behind the change in seat back angle. In contrast, the
thorax moved approximately at the same rate as the seat
back (coefficient 0.9), so that lumbar spine flexion decreased
with increasing recline. Over the seat back angle change of
30, lumbar spine flexion decreased an average of 17.
Consistent with previous studies (e.g., Reed et al. 2013;
Park et al. 2018), the lap belt was typically above and for-
ward of the pelvis by approximately a belt width on average,
with the belt further forward and higher for individuals with
higher BMI. Lap belt fit changed slightly with increased
recline, with the belt shifting rearward, closer to the pelvis.
However, because the vertical position with respect to the
ASIS remained the same, on average, the effect that this
change would have on lap belt performance in crash scen-
arios is unclear. Park et al. (2018) showed that the lap belt fit
placement on passengers is typically much farther from the
pelvis than is the case with anthropomorphic test devices.
The participants placed the shoulder belt significantly fur-
ther inboard (closer to occupant centerline) at higher recline
angles, to the extent that at 53most participants placed the
belt over the top of the sternum (suprasternale landmark)
rather than centered on the clavicle. Note that the participants
donned the belt after finding a comfortable reclined posture,
which may be a different sequence from what would be
expected in a vehicle. The D-ring (upper anchorage) location
was rotated around the seat H-point at the same rate as the
seat back, creating nominally constant belt geometry. More
study is needed to understand why the belt fit changed in this
way, whether this change is realistic for long-duration sitting,
and the potential safety consequences.
This study has important limitations that should be
addressed in future work. A single seat was used, and this seat
was not designed for highly reclined postures. Seats specific-
ally designed for such postures may have different contours
and kinematics and could produce different postures. For
example, a different seat contour may produce different effects
of seat back angle on lumbar spine flexion. The seat belt was
not designed for reclined postures and the particular contour
of the seat many have interacted with body shapes in ways
that influenced the findings with respect to the effects of body
size on posture. Only an approximation of a seat-integrated
belt was presented. Posture measurements were made during
a short-duration sitting session. The observed postures were
largely sagittally symmetric, but long-duration, reclined sitting
might result in increased asymmetry. The data were also gath-
ered in a static laboratory setting. Vehicle ride motion could
result in differences in posture. The findings are also limited
by the characteristics of the subject pool. A larger, more
diverse subject pool would allow more powerful tests of the
effects of age and body characteristics and particularly the
potential interactions among age, sex, and body dimensions.
References
Lin H, Gepner B, Wu T, Forman J, Panzer M. 2018. Effect of seatback
recline on occupant model response in frontal crashes. Proceedings
IRCOBI Conference; Athens, Greece. http://ircobi.org/
Park J, Ebert SM, Reed MP, Hallman JJ. 2015. Development of an opti-
mization method for locating the pelvis in an automobile seat.
Paper presented at: 6th International Conference on Applied
Human Factors and Ergonomics, Las Vegas, Nevada.
Park J, Ebert SM, Reed MP, Hallman JJ. 2016a. A statistical model
including age to predict passenger postures in the rear seats of auto-
mobiles. Ergonomics. 59(6):796–805. 10.1080/00140139.2015.1088076
Park J, Ebert SM, Reed MP, Hallman JJ. 2016b. Statistical models for pre-
dicting automobile driving postures for men and women including
effects of age. Hum Factors. 58(2):261–78. 10.1177/0018720815610249
Park J, Ebert SM, Reed MP, Hallman JJ. 2018. Comparison of three-
point belt fit between humans and ATDs in rear seats. Traffic Inj
Prev. 19(sup1):S65–S9.
Reed MP, Ebert SM, Hallman JJ. 2013. Effects of driver characteristics
on seat belt fit. Stapp Car Crash J. 57:43–57.
Reed MP, Manary MA, Flannagan CAC, Schneider LW. 2002. A statis-
tical method for predicting automobile driving posture. Hum
Factors. 44(4):557–68.
SAE International. 2018a. J826: Devices for use in defining and measuring
vehicle seating accommodation. Warrendale (PA): SAE International.
SAE International. 2018b. J1100: motor vehicle dimensions.
Warrendale (PA): SAE International.
Figure 5. Distribution of lap belt locations relative to estimated bone ASIS on
the right side of the pelvis. Condition means and standard deviations are
shown. Data from smaller back angles are shown with smaller symbols. For ref-
erence, the belt webbing width is 45 mm.
Figure 6. Torso belt score as a function of seat back angle. Higher scores are
more outboard. A score of zero indicates that the inner edge of the belt lies on
the body centerline at the height of the clavicle (suprasternale landmark).
S42 M. P. REED ET AL.