Abstract
Background There is conflicting evidence on the association between sagittal neck posture and neck pain.
Objective The purposes of this study were: (1) to determine the existence of clusters of neck posture in a cohort of 17-year-olds and (2) to establish whether identified subgroups were associated with biopsychosocial factors and neck pain.
Design This was a cross-sectional study.
Methods The adolescents (N=1,108) underwent 2-dimensional photographic postural assessment in a sitting position. One distance and 4 angular measurements of the head, neck, and thorax were calculated from photo-reflective markers placed on bony landmarks. Subgroups of sagittal sitting neck posture were determined by cluster analysis. Height and weight were measured, and lifestyle and psychological factors, neck pain, and headache were assessed by questionnaire. The associations among posture subgroups, neck pain, and other factors were evaluated using logistic regression.
Results Four distinct clusters of sitting neck posture were identified: upright, intermediate, slumped thorax/forward head, and erect thorax/forward head. Significant associations between cluster and sex, weight, and height were found. Participants classified as having slumped thorax/forward head posture were at higher odds of mild, moderate, or severe depression. Participants classified as having upright posture exercised more frequently. There was no significant difference in the odds of neck pain or headache across the clusters.
Limitations The results are specific to 17-year-olds and may not be applicable to adults.
Conclusion Meaningful sagittal sitting neck posture clusters were identified in 17-year-olds who demonstrated some differences with biopsychosocial profiling. The finding of no association between cluster membership and neck pain and headaches challenges widely held beliefs about the role of posture in adolescent neck pain.
Neck pain is the fourth most common health disorder for years lived with disability internationally.1 Worldwide, the percentage of health care resources spent on neck pain are rapidly increasing.2 Neck pain is reported in childhood3 and rises in prevalence throughout adolescence.4–7 By 18 years of age, neck pain prevalence is equivalent to adult levels.5 Neck pain disorders are more common in females4,7–10 and have been associated with many different biopsychosocial factors. Associated anthropometric factors include small or tall height.11 Lifestyle factors include increased computer use12 and increased time spent sitting.13 Psychological factors include depression,14 poor mental health,10,15 and stress and psychological distress.10,15 Other factors are comorbid pain sites (eg, back pain),7,10,15 ill health,7,15 and physical factors such as reduced muscle endurance16 and altered spinal posture.17–22
Neck posture, particularly in a sitting position, is considered an important contributing factor to the development and persistence of neck pain18–23 and headaches.24 Consistent with this fact, neck posture is commonly targeted by clinicians to prevent and reduce neck pain. Sagittal neck posture in a sitting position has been investigated previously due to a common clinical belief that a more forward head posture (FHP) (where the head is anterior to the trunk in the sagittal plane)19,25 is associated with neck pain and headaches.26 Forward head posture may increase load on passive cervical structures (joints, ligaments)25,27 and on the posterior neck musculature by the increased gravitational moment28,29 and torque around C7.30 This increased load is manifest by increased extensor muscle activity demonstrated in FHP.31,32
Although these mechanisms are plausible, the link between neck posture and neck pain is unclear.33 Some studies18,20,21,34 have shown an association between neck posture and neck pain, whereas other studies23,35–39 have not. Factors contributing to the discordant findings include use of different angular or translational posture variables, inclusion or exclusion of thoracic measures and upper cervical spine measures/head tilt, whether the neck posture was assessed in a sitting or standing position, the task during which posture was assessed, definitions of neck pain, sample size, and demographics. These methodological differences make direct comparisons between studies difficult.
Another potential reason for the disparity in the literature in determining the association between posture and neck pain could be the existence of neck posture subgroups. Subgroups of sagittal lumbar spine standing posture have been identified using sagittal photography in an adolescent population40 and sagittal radiographs in adults.41 In adolescents, individuals with “nonneutral” standing lumbar posture had higher odds for back pain.40 In adults, nonneutral postures were more commonly associated with symptomatic back pain or degenerative disease.41 Cluster modeling appears highly suited to the examination of posture subgroups,40 but this approach has not been applied to investigate associations between sagittal neck posture and neck pain.
The literature supports the supposition of subgroups of sitting neck posture. Forward head posture can occur via a combination of upper thoracic flexion and cervical spine flexion.25,42 Two versions of FHP have been described, one with and one without extension of the upper cervical spine.43 Alternatively, FHP may exist in moderate and extreme forms.25,43 Another possible subgroup could be the opposite extreme of FHP, a very upright position with the head held close to the vertical axis.34 Although there is a lack of consensus as to what constitutes “neutral” neck posture,44 it is likely that this fourth subgroup exists, between the 2 posture extremes. Thus, the first aim of this study was to determine whether sagittal sitting neck posture subgroups that correspond to potential variants described in the literature could be identified with cluster modeling in late adolescence. The second aim was to profile identified posture clusters based on factors potentially associated with sitting posture (sex, anthropometrics, lifestyle, psychological factors). The final aim was to investigate the association between identified neck posture clusters and neck pain and headaches. Significant insight into the potential association between neck posture and neck pain could be achieved by applying cluster analysis to neck posture data.
Method
Study Design
This was a cross-sectional study in the Western Australia Pregnancy Cohort (Raine) study (http://www.rainestudy.org.au).
Participants
The Raine study commenced in 1989, with 2,868 children born to 2,804 mothers. Ethnicity was predominantly Caucasian. At 17 years of age, demographic characteristics of the participating families were similar to the Western Australian population for families with 15- to 17-year-old children, except for a lower proportion of rural-dwelling families and a higher proportion of families dwelling in high socioeconomic areas.45
The 17-year-old cohort follow-up took place between June 2006 and December 2009. Participants (average age=17.0, SD=0.2) completed a questionnaire that included a description of neck pain and headache and a variety of psychosocial domains. Participants underwent physical testing for height, weight, and photographic sagittal posture assessment. Consent was obtained from the participants.
Measures
Postural assessment.
Postural assessment methods were the same as those used in a prior reliability study that demonstrated reliability and practicality of use in large-scale studies of adolescent posture.46 For consistency, photographic reflective markers were placed on bony landmarks on the participant's right side. Landmarks utilized were the outer canthus, tragus, and C7 and T12 spinous processes. A 25-cm plumb line was hung from the stool to calibrate distance and determine vertical. An Olympus camera (Olympus FE-130, Tokyo, Japan) was placed on a tripod 80 cm from the floor and 250 cm perpendicular to the participant. Right-sided lateral photographs were taken as participants sat on a stool with thighs horizontal and knees flexed to 90 degrees. The standardized instructions used to position the participants in order to obtain the usual posture of the head and neck were the same as those used in a previously reported reliability study.46 Participants were instructed to “put your hands about two-thirds of the way down your thighs with the palms up, sit like normal and relax, and look straight ahead.” Markers were digitized using Peak Motus motion analysis system version 8 software (Peak Performance Technologies Inc, Centennial, Colorado). Measures (Figure) were selected to best represent the attributes of sagittal posture that relate to the proposed mechanism for neck pain25 and the components of posture routinely assessed in clinical practice.47
Angular and distance measures describing sitting neck posture alignment.
Interrater reliability has been reported as fair to good for these posture variables.46 The reliability of intrarater redigitization for all distance and angular measures used in this study were found to be excellent in this cohort at age 14 years.46 Radiographic examination is considered the gold standard for postural measurement.48 However, its disadvantages include exposure of individuals to ionizing radiation and high cost, making it unsuitable for use in large population studies. Photographs, although not as accurate as radiographs, can detect meaningful classifications of lumbar spine posture that correspond to x-ray classifications,40 and their use is supported in large population studies.46
Anthropometrics.
Body mass index (BMI) was calculated from height and weight measures using standard procedures.
Lifestyle factors.
For physical activity participants were asked: “Outside school, Technical and Further Education (TAFE), or work hours, how often do you usually exercise in your free time, so much that you get out of breath or sweat (once a month or less/once a week/2–3 times a week/4–6 times a week/every day)?” A binary variable was made to separate those participants who exercised once a week or less from those who exercised more frequently than once a week. For sedentary behavior, participants were asked about their computer use and time spent sitting. For computer use, they were asked: “On average, how many hours per day do you usually use a computer (eg for school/work, games, Internet) on a weekday (not at all/less than 1 hour/about 1–2 hours/about 2–4 hours/more than 4 hours)?” A binary variable was used to split those participants with low or no use (<5 hours per week on a weekday) from those using a computer more regularly. Time spent sitting was taken from each participant's response to the question “During the last 7 days, how much time did you spend sitting on a weekday (weekend day)?” on the International Physical Activity Questionnaire (IPAQ).49 From these responses, a weighted sum variable was calculated representing total hours sitting per week.
Psychological factors.
Depressed mood was assessed using the 20-item Beck Depression Inventory for Youth (BDI-Y).50 Participants were asked to rate the frequency of their symptoms over the preceding 2 weeks on a scale of 0 (never) to 3 (always). Responses were summed to produce a total score ranging from 0 to 60. The BDI-Y has good reliability.50 Scores were categorized into a binary variable of “no or minimal depression” versus “mild, moderate, or severe depression.”51
Neck pain and headache.
Participants were asked to look at a picture depicting neck pain as pain in the neck and upper trapezius muscle region and to answer the follow questions: (1) “Have you ever had neck/shoulder pain (yes or no)?” (2) “Has your neck/shoulder pain lasted more than 3 months continuously (it hurt more or less every day) (yes or no)?” (3) “Has your neck/shoulder pain ever lasted for more than 3 months off and on (it hurt at least once a week, but not every day) (yes or no)?” and (4) “Has sitting ever made your neck/shoulder pain worse (yes or no)?” These questions were adapted from the Nordic questionnaire, which has established validity and reliability.52 Positive responses to the first 3 questions were converted to a new variable (ie, presence of persistent neck pain) to examine neck pain of recurrent and prolonged nature. For headaches, participants were asked, “Do you have now, or have you had in the past, any of the following health professional–diagnosed medical conditions or health problems: migraine or severe headache (yes or no)?”
Data Analysis
Initial data screening was conducted using boxplots and identified 20 outliers (values 1.5 times the interquartile range). Of these outliers, 4 participants' photographs were unusable due to quality issues. For the remaining 16 outliers, an error was found in the digitization of 2 markers used to form the distance calibration. This error meant that when the data was used to normalize the scale from those participants, the distance measure (ie, head displacement) was incorrect. To correct the digitization of the markers on the distance calibration device, a correct scaling factor was applied to head displacement measures.
Prior to clustering, the correlation matrix was examined to identify unduly high correlations between posture variables that might result in overrepresentation in the cluster solutions.53 A 2-step cluster analysis was performed using standardized (z scores) posture measures. A hierarchical technique, Ward's linkage, was used to derive the number of clusters and their cluster centers, followed by the nonhierarchical, iterative technique of K-means cluster analysis using the cluster seeds generated from the Ward's linkage solution.
Validity of the cluster solution was assessed in 3 ways. First, it is recommended that the determination of an optimal cluster solution be performed in conjunction with consideration for the interpretability and theoretical meaning of the generated clusters.53 Determination of the number of clusters to be estimated was guided by prior literature regarding subgroups of neck posture and the degree to which estimated subgroups corresponded to these subgroups. As prior literature suggested that there may be, in addition to a “neutral” head posture, 2 different FHPs (with and without upward head tilt),25,42 a 3-cluster solution was estimated. Additionally, given the potential of a very upright posture,34 a 4-cluster solution was estimated. Clusters from 3- and 4-cluster solutions were compared with posture subgroups previously described in the literature, and photographs of individuals with values near the centroid of the clusters were examined to interpret their potential clinical relevance. Second, split-half sampling of the final 4-cluster solution was conducted. Third, criterion validity of the 4-cluster solution was evaluated by examining profiles across variables not used to form the clusters that had theoretical support for variation across clusters.
Profiles of clusters across variables of interest were assessed using logistic regression (binary variables), ordinal logistic regression (ordinal variables), or linear regression (continuous variables), adjusting for sex due to sex differences across clusters. Sex interactions with profile variables also were assessed to investigate whether patterns of cluster difference varied by sex. The Brant test was used to check the proportional odds assumption for ordinal logistic regression. Multivariable logistic regression was used to estimate associations among neck pain, headache, and posture clusters after adjusting for the potential confounding variables.
Analysis was performed with Stata version 13 for Mac (StataCorp LP, College Station, Texas). Statistical significance was set at α=.05.
Role of the Funding Source
The authors acknowledge National Health and Medical Research Council (NHMRC) program grant 353514 and NHMRC project grant 323200 and additional funding for core management from The University of Western Australia (UWA); Raine Medical Research Foundation; Telethon Kids Institute; UWA Faculty of Medicine, Dentistry and Health Sciences; Women and Infants Research Foundation; Curtin University; and Edith Cowan University. Dr Beales and Dr Straker were supported by research fellowships from the NHMRC of Australia.
Results
Of the 1,475 participants in the 17-year-old cohort follow-up, 1,123 underwent physical examination, including height, weight, and posture measurements. Following data inspection and cleaning, 1,108 participants had complete posture measurements available for clustering.
Cluster Analysis
Following conversion of postural data to standardized z scores, neck flexion and head flexion had a strong negative correlation (r=−.73). To avoid overrepresentation in the cluster solution, head flexion was not used, as it had been shown to have a lower interclass correlation coefficient for consistency and absolute agreement compared with neck flexion.46
Ward's linkage analysis determined the cluster centers of the 5 remaining postural measures (thoracic flexion, neck flexion, cervicothoracic angle, craniocervical angle, and head displacement). Examination of the agglomeration schedule and stopping rules (Calinski-Harabasz pseudo_F index and Duda-Hart index) provided most support for the 3- or 4-cluster solutions. Therefore, subsequent K-means cluster analysis estimated both 3- and 4-cluster solutions.
The 4-cluster solution split one large cluster from the 3-cluster solution into 2 clusters. These new clusters exhibited similar profiles; however, on photographic inspection, one group demonstrated less thoracic flexion and appeared to sit closer to end-range thoracic extension compared with the subgroup it had split from. This biomechanical difference was deemed to be of potential clinical importance, as previous research has demonstrated that individuals at extremes of range were at greater risk of neck pain.34 Therefore, the 4-cluster solution was chosen as the most supported by the data and clinical relevance. Split-half sampling confirmed very similar 4-cluster solutions.
Cluster Descriptions
Table 1 presents cluster postural measures.
Mean (Standard Deviation) Postural Measures for Each Cluster With Example Participants From Each Clustera
Cluster 1 (n=311, 28%, upright) was characterized by the least thoracic flexion and neck flexion of all clusters. There was a small cervicothoracic angle, and the craniocervical angle was the smallest of all clusters (ie, less upward head tilt). The head was displaced the smallest distance forward compared with all other clusters.
Cluster 2 (n=265, 24%, intermediate) demonstrated a similar pattern of characteristics as cluster 1, but had greater thoracic flexion. When considering the position of the neck on the thorax, the cervicothoracic angle was greatest of all of the clusters, indicating that this subgroup was the least flexed. There was less neck flexion, less forward head displacement, and slightly smaller craniocervical angle than the overall mean.
Cluster 3 (n=178, 16%, slumped thorax/forward head) was characterized by the most thoracic flexion and the most neck flexion of all clusters. The head was displaced the greatest distance forward, and the craniocervical angle was greater than the other clusters (ie, tilting the head upward).
Cluster 4 (n=354, 32%, erect thorax/forward head) was characterized by slightly less thoracic flexion than the population mean (but more thoracic flexion than cluster 1) and more neck flexion than the group mean. This cluster had the smallest cervicothoracic angle. The head was displaced forward (more than clusters 1 and 2 and a similar distance to cluster 3) with a slightly increased craniocervical angle (ie, tilting the head upward) (Tab. 2).
Descriptive Statistics of Study Measures by Sex and Overalla
Cluster Characterization
Clusters 1 (upright) and 4 (erect thorax/forward head) had higher proportions of female participants than male participants (P<.001, Tab. 3). Subsequent analyses were adjusted for sex, and the absence of sex interactions was confirmed for all analyses (ie, no significantly different patterns of difference across clusters in male versus female participants).
Profiles of Anthropometric, Lifestyle, and Psychological Factors Across Clusters and Sex-Adjusted Cluster Differencesa
Anthropometrics.
Participants in cluster 1 (upright) were significantly taller than those in the other clusters (Tab. 3, P=.011). Participants in cluster 3 (slumped thorax/forward head) and cluster 4 (erect thorax/forward head) were significantly heavier than those in cluster 1 (upright) and cluster 2 (intermediate) (P<.001). The same association was seen between cluster membership and BMI (P<.001, Tab. 3).
Lifestyle factors.
Participants in cluster 1 exercised more than those in the other clusters (P=<.001) (Tab. 3). There were no differences among clusters for the amount of time spent sitting (P=.452) or using the computer (on weekdays) (P=.356).
Psychological factors.
Two hundred eighty-two participants (23%) were classified as having mild or more depression with the BDI-Y (Tab. 2). After adjusting for sex, cluster 3 (slumped thorax/forward head) had greater odds for depression (P=.047, Tab. 3).
Neck pain and headache.
Two hundred nineteen participants (22%) reported persistent neck pain (Tab. 2), with 141 (64%) being female. After adjusting for sex, there was no difference among the clusters and the odds of persistent neck pain (P=.773, Tab. 4). To determine whether any association was masked by other variables shown to be different among clusters, further adjustment for sex, height, weight, exercise frequency, and depression was undertaken. There was still no significant difference between the clusters and the presence of persistent neck pain following this (P=.741, Tab. 4). One hundred forty participants (14%) reported neck pain worse with sitting (Tab. 2). After adjusting for sex or a multivariable model of sex, height, weight, exercise frequency, and depression, the odds of neck pain made worse by sitting did not differ among the clusters (P=.150 and P=.262, respectively, Tab. 4).
Profiles of Neck Pain and Headache Across Clusters and Adjusted Cluster Differencesa
Ninety-six participants (9%) reported headaches (Tab. 2), with 51 (53%) being female. After adjusting for sex or a multivariable model of sex, height, weight, exercise frequency, and depression, no significant difference was found between clusters and headaches (P=.563 and P=.450, respectively, Tab. 4).
Discussion
Cluster analysis identified 4 subgroups of sitting neck posture. All were well characterized by multiple measures and their validity was supported by previous research. Other strengths included utilization of a large, community-based cohort of similar age and consideration of differences in profiles of biopsychosocial factors.
Sitting Neck Posture Clusters
Previous research has described FHP as a posture variant.25,33,43,54 These results support that it is common for 17-year-olds to hold their head anterior to their trunk in the sagittal plane. Two clusters characterized by a FHP were identified (Tab. 1) and were separated by the amount of thoracic flexion. Cluster 3 was characterized by the greatest amount of thoracic flexion of all of the clusters, and cluster 4 was characterized by a more erect thorax. The FHP of cluster 3 (slumped thorax/forward head) supports the anecdotal descriptions of FHP and a previous study where thoracic flexion in a sitting position was associated with a forward lean of the neck.25,55
There was no clear prior report of an erect thorax in association with FHP in a sitting position (cluster 4). This finding may be explained by the majority of previous studies utilizing singular postural measures in the neck or not having considered the thoracic spine position. One study demonstrated that FHP in a standing position was not associated with increased upper or lower thoracic curvature.26 Differences in postural control strategies and thoracic position may occur between sitting and standing, potentially accounting for differing results. Distinctly different thoracic flexion in clusters 3 and 4 supports inclusion of this angle in future studies investigating sitting neck posture.
Another difference between clusters 3 (slumped thorax/forward head) and 4 (erect thorax/forward head) was a larger craniocervical angle in cluster 3, indicative of a greater degree of upward head tilt. This difference is consistent with previous FHP descriptions42,56 and previous neck posture measurements.25,55 A potential role of the head in neck posture is to level the eyes.25,44 Cluster 3 (slumped thorax/forward head) aligns to this, with greater thorax flexion but also more upward head tilt and thus a level eye gaze. This finding contrasts with a study conducted in a standing position that indicated a more flexed neck was not associated with a concomitant tilt of the head upward.26 The lack of agreement may be explained by the relationships of adjacent segments altering in a sitting position.55
In contrast to the FHP of clusters 3 and 4, cluster 1 (upright) had the most upright sitting posture (Tab. 1). This finding is consistent with a previous study that showed decreased forward lean of the neck was associated with decreased flexion in the lower cervical spine and increased flexion of the upper cervical spine was associated with a downward tilt of the head.55 It makes sense that if both the thorax and neck are less flexed, the head will require less upward tilt to level the eyes, as evidenced by the smaller craniocervical angle in cluster 1.
Cluster 2 posture (intermediate) is somewhat in the middle of the 2 extreme postures in this study (cluster 1 [upright] and cluster 3 [slumped thorax/forward head]). Although there is little evidence to determine what constitutes a neutral posture,44 the intermediate posture of cluster 2 appears to be consistent with the mid-range position of the neck and thorax and with the head in a position where the eye gaze is level.
Posture Clusters, Sex, and Anthropometric Factors
Similar to previous findings35,36,57,58 female participants were more commonly represented in clusters with more upright sitting postures. It is not clear why female individuals sit more upright, but it could relate to sex differences in other anthropometric factors and social factors related to body image.58 Body mass index can influence sitting posture,59 with a higher BMI associated with more slump in a sitting position.58 However, neither of these studies58,59 considered the neck and head position. Results from this study concur with a prior finding that high school students who are overweight sit in more neck and thoracic flexion.34 This increased neck and thoracic flexion may be a consequence of increased load associated with higher BMI or comorbid muscle deconditioning making it more difficult to hold more erect postures.60
Posture Clusters and Psychosocial Factors
Cluster 3 participants (slumped thorax/forward head) had higher odds of depressive symptoms (Tab. 3), consistent with a previous study in adolescents with chronic nonspecific musculoskeletal pain for whom slumped postures were associated with higher levels of anxiety and depression.61 Of note, slumped lumbar sitting posture was not significantly associated with higher scores of depression as measured with the BDI-Y in the Raine study cohort at the age of 14 years.58 This difference is likely due to the head and neck position in the current modeling, but only trunk position in the previous study.58 A sharp rise in the prevalence of depression in adolescents following the onset of puberty62 could indicate that depression would be a greater influence at 17 years of age than at 14 years of age. An association between cluster membership and depression supports the concept of a mind-body relationship, although the direction of this relationship is unknown.58,63,64
Posture Clusters and Lifestyle Factors
Cluster 1 participants (upright) were more physically active compared with the other clusters. Physical activity can relate to greater back muscle endurance, which has an association with upright posture.65 This conclusion is consistent with an intervention study showing exercise can facilitate adoption of more upright postures.66
Computer use did not differ among clusters. A previous study67 showed transient changes in neck and head tilt posture with use of different types of information technology. Higher computer use was associated with increased head flexion and neck flexion in male participants but not female participants in the Raine study cohort at age 14 years.12 Although Straker et al12 only used singular postural measures, results from the current study do appear contrary, as there was no increased computer use in the 2 FHP clusters. Increased use of computers, mobile touch screen tablets, and smartphone devices since this study was undertaken may mean that exposure time has increased and may be relevant to neck posture in adolescents now.
Posture Clusters and Pain
There is a strong clinical and societal belief that neck posture is a significant factor associated with neck pain.68 There is some supporting evidence for this view,17,19 although other research refutes it.35–39 The 4 neck posture clusters were not associated with persistent neck pain or neck pain made worse by sitting (Tab. 4), despite accounting for limitations in prior research, including use of well-defined measures of posture and good characterization of pain. In this study, participants with 3 months of recurrent or prolonged neck pain were included for analysis, whereas previous studies have not specified the duration of neck pain.18,20,39 This study also considered neck pain to be more specific to the task of sitting and to align with the test position used in an attempt to replicate clinical practice of matching aggravating activities to specific objective findings. To our knowledge, no previous research has detailed potential relationships between neck posture and neck pain that is worse in a sitting position.
Previous studies indicating an association between neck posture and neck pain have drawn participants from clinical populations18,39 or workplaces that involve prolonged computer use, an occupational task reported to link with neck pain perhaps as consequence of the prolonged neck flexion postures adopted.20 Smaller, specialized samples may misrepresent associations in the general population.
The current results do not support the commonly held clinical and societal belief that neck pain is related to spinal posture. This finding is consistent with findings from systematic reviews that the association between neck pain and posture is weak.22,33 In contrast, previous studies showed that factors such as genetics,69 female sex,4,7–10 depressed mood, stress,14 and sleep patterns70 are associated with neck pain. These findings suggest that neck pain is associated with changes in pain regulatory mechanisms rather than biomechanics, which supports calls to consider and manage neck pain from a broader biopsychosocial perspective.
There was no difference in the odds of headaches across the 4 clusters, but the definition of headache used in this study was broad. A previous study showed that neither cervicogenic headache nor migraine was associated with head and neck posture.71 Most previous studies looking for a link between neck posture and headaches have focused on cervicogenic headache24 or tension headache groups.72,73 Even when considering headache type separately, results are contradictory, with some studies reporting an association between posture and headaches,24,72 whereas other studies did not.71,73 These studies used single posture angle measures. More research is warranted using posture clusters and clearly defined headache groups (thus avoiding any washout effect from subgroups of neck posture or subgroups of headaches).
Although there was not a cross-sectional association between posture clusters and neck pain, prospective examination of the effect of cluster membership on future episodes of neck pain is worth consideration. Flexed neck posture was not associated with incident neck pain over 3 years in one large prospective study of workers in a wide range of work roles across 34 workplaces.13 The lack of association observed may have been due to limitations of analysis being performed on mean values from subsamples of the workers at each workplace rather than at an individual level.
Limitations
Posture clusters and experience of pain were not associated in this study. The adolescent age of the participants limits the extrapolation of results to adults. Furthermore, participants in this study were primarily Caucasian Australians. Associations among posture, anthropometric measures, psychological factors, lifestyle factors, and neck pain may be different in populations from other countries or other ethnicities. Assessment of standardized posture with 2-dimensional photography cannot capture the complex interactions of intersegmental translation and rotation that may be measured by radiographic assessment.48 Another limitation is that photographs represent only a “snapshot of time” and cannot reflect changes in posture that occur over time, which may affect potential relationships between sitting posture and neck pain. Complex interactions among sitting posture, sitting task, pattern of breaks, and other contributing factors were not captured. Although the lumbar spine posture was not included in the modeling, we acknowledge that it may be important.
In conclusion, this study identified 4 neck posture subgroups in a sitting posture. The results support previous findings that sitting posture is associated with different dimensions, including exercise frequency, depression, and BMI.34 Despite strong support for the existence of neck posture subgroups, they were not associated with persistent neck pain, neck pain in a sitting position, or headaches in 17-year-olds. This finding raises questions regarding the efficacy of generic postural advice for adolescents with and without neck pain.
Footnotes
All authors provided concept/idea/research design and writing. Dr O'Sullivan and Dr Straker provided data collection and project management. Ms Richards, Dr Beales, Dr Smith, and Dr Straker provided data analysis. Dr Straker provided fund procurement, facilities/equipment, and institutional liaisons. The authors thank the Raine study participants and their families, and the Raine study team for cohort coordination and data collection.
Ethical approval for the study was obtained from the Curtin University Human Research Ethics Committee (Reference HR 84/2005) and the Princess Margaret Hospital Human Research Ethics Committee (Reference 1214EP).
The authors acknowledge National Health and Medical Research Council (NHMRC) program grant 353514 and NHMRC project grant 323200 and additional funding for core management from The University of Western Australia (UWA); Raine Medical Research Foundation; Telethon Kids Institute; UWA Faculty of Medicine, Dentistry and Health Sciences; Women and Infants Research Foundation; Curtin University; and Edith Cowan University. Dr Beales and Dr Straker were supported by research fellowships from the NHMRC of Australia.
- Received December 2, 2015.
- Accepted April 14, 2016.
- © 2016 American Physical Therapy Association