Research Review By Dr. Kent Stuber©

Date Posted:

January 2010

Study Title:

Pillow use: the behaviour of cervical pain, sleep quality and pillow comfort in side sleepers


Gordon SJ, Grimmer-Somers K, Trott P

Author's Affiliations:

School of Public Health, Tropical Medicine, and Rehabilitation Sciences, James Cook University; School of Health Sciences, University of South Australia

Publication Information:

Manual Therapy 2009; 14: 671-678.

Background Information:

Many of our patients ask us about pillows. Unfortunately, there has been little research to date on pillow usage and neck pain beyond a few small projects comparing different pillow shapes, resulting in insufficient evidence of suitable quality to allow for definitive conclusions.

Thus it is difficult for clinicians to make evidence informed recommendations to their patients with respect to pillow types and shapes. Clinicians have to rely on what they consider common sense (“don’t sleep with three pillows under your head”), information from pillow manufacturers, or other forms of anecdotal evidence upon which they must base their recommendations.

Pertinent Results:

These authors began their research with a telephone survey of 800 people that found that 72% of their respondents were side sleepers. Types of pillows used by these respondents included polyester (45%), rubber (14.4%), foam contour (12%), feather (9%), and foam regular shaped (8%). 18% of their subjects woke at least once weekly with neck pain. This formed the basis for the larger portion of the study.

In the RCT portion, 106 subjects participated (33 were male, participant ages ranged from 20-81 with an average of just over 49 years). Each subject tested a variety of pillows individually. 105 subjects participated in the polyester pillow trial (95.4% competed it) compared with 101 in the foam regular pillow trial (94.1% completed), 103 in the foam contour pillow trial (88.3% completed), 101 in the feather pillow trial (67.3% completed), and 100 in the rubber pillow trial (97.0% completed).

Pertinent Results Include:
  • There were 7 dropouts from the trial at various stages.
  • 30% of the subjects had no waking pain at all during the trial on any pillow type.
  • Women under 40 years old reported significantly more waking pain events on average than all other groups.
  • Subjects’ own pillows and feather pillows were responsible for the most reports of known reasons for waking neck pain. Regular-shaped foam pillows and feather pillows most often produced events of waking pain, compared with the polyester, foam contour, and rubber pillows which were more likely to continue patterns of no waking pain.
  • Rubber pillows were best at ensuring that subjects who went to sleep without pain awoke without pain while the feather pillow was the worst at this.
  • Pillow shape (regular foam vs. contoured foam) appeared to have little effect on temporal pain patterns.
  • When compared with subjects’ own pillows and changing pain patterns 10.4% were worse when using the polyester pillows, 12.5% were improved; 16.9% were worse with the foam regular pillows while 22.7% were improved; 9.3% were worse with the foam contour pillows while 11.1% were improved, 16.0% were worse with the feather pillows while 22% were improved, and 6% were worse with the rubber pillows while 12% were improved with the rubber pillows.
  • When compared to usual pillows, subjects were significantly more likely to report low pillow comfort with all of the different trial pillows (with odds ratios of 1.4 for the polyester pillow, 1.5 for the foam regular pillow, 2.4 for the foam contour pillow, and 8.4 for the feather pillow) except the rubber pillow which had a protective odds ratio of 0.5.
  • When compared with usual pillows, foam contour and feather pillows were significantly more likely to produce low sleep quality with odds ratios of 1.5 and 3.7 respectively; the rubber pillow was significantly protective against low sleep quality with an odds ratio of 0.6.
  • Comfort and quality ratings were significantly lower for feather pillows (41%/40%) when compared with usual pillows (85%/71%), polyester (81%/71%), foam regular (80%/72%), and rubber (91.5%/81%).

Clinical Application & Conclusions:

This study breaks new ground as it provides an initial evidence base that clinicians can take to patients in recommending pillows to aid them with neck pain issues, particularly waking neck pain, as well as sleep quality and pillow comfort. The findings seem to contradict many of the anecdotally based recommendations frequently made by clinicians and manufacturers. Pillow shape did not affect pain levels, and contoured pillows were found to be less comfortable and provide worse quality sleep than regularly shaped pillows. Patients’ regular pillows performed as well as foam and polyester pillows. Feather pillows performed poorly in all respects and the authors concluded that they should not be recommended. Rubber pillows outperformed subjects’ own pillows in all outcome measures and the authors felt that these could be recommended to patients.

As a clinician one of my immediate questions was what type of rubber pillow should I be recommending? In this study a Dunlopillo latex rubber pillow was employed with a measurement of 115mm in depth (height). A quick glance online finds that Dunlopillo is a UK based company and thus North American clinicians may want to look into more local product lines to recommend to their patients.

Study Methods:

The study was a 10 week random allocation block design field study that involved subjects trying 5 different (test) pillows in addition to their own. Subjects tried the following pillow types:
  1. Polyester synthetic fibre filled
  2. Foam regular shape (from high density foam)
  3. Foam contour shape (high density foam)
  4. Latex rubber
  5. Feather
The authors used a convenience sample (from previous studies) along with subjects recruited via newspaper. Subjects had to be adults, side sleepers, and without a current neck pain complaint. Subjects were randomized into 1 of 5 blocks, each of which received a different pillow (different pillow allocation was then rotated between the blocks). Subjects completed a 1 week diary for each pillow, where they were to indicate retiring and waking neck pain ratings, the duration of waking neck pain, pillow comfort, and sleep quality ratings.

Pain ratings were classified according to temporal patterns (4 possibilities), and a cumulative waking pain frequency-duration score for each pillow’s week was calculated. These waking pain cumulative scores were put into 6 different categories.

In week 1, subjects used their own pillows (this was considered the gold standard for comparison); subjects spent the next 9 weeks alternating between one of the trial pillows (according to the block to which they were allocated) for one week and their own pillow for one week (which was to allow for a washout). If subjects felt that one of the trial pillows was causing them harm (loss of sleep or causing other symptoms) they were to discontinue use of that particular pillow (this data would be considered among the drop-out data).

The authors used descriptive analysis, as well as a survival analysis to deal with drop outs, tests of significance (chi square with significance set at .05), ANOVA modeling, and calculated odds ratios (using 95% confidence intervals) for waking pain between pillow pairs, comparing the pillow in question with a subject’s own pillow. Comparisons were also made between the different test pillows. Multivariate logistic regressions were also calculated to test associations between pillows and pain categories and for confounding effects of age and gender.

Study Strengths / Weaknesses:

The authors employed a unique set of methods that had its strengths and weaknesses. Having subjects rotate through each different pillow with a washout period in between is a strength as it turns the study into a large cross-over trial. Using the washout period with subjects using their own pillow was an intelligent addition to the study design.

The authors felt that the 1 week trial period for each pillow was suitable, but a longer trial period (at least one month) would be more desirable as it would allow them to see if more dropouts occurred over time and would allow for a more complete picture to arise. Patients typically use the same pillow for a period of months to years and thus the 1 week period used here may not have been adequate.

The authors did not take anthropometric measurements to see if certain pillow heights, shapes, etc might have been better suited to subjects with different sized frames, widths or shoulders, etc and this may be beneficial in the future.

The authors note that subject blinding to pillow type was largely impossible and there was a dependence on subject self-reporting which could call the reliability of the results into question. However, this is a problem inherent in many field trials. The authors also did not record data on the subjects own pillows (type, shape, age, etc) and that may have been beneficial in subsequent comparisons and analysis.

It is also important to keep in mind that only 1 type of foam regular pillow, feather pillow, foam contoured pillow, polyester pillow, and rubber pillow were investigated – there are many other different types of pillows and different shapes as well, so how they compared with those in the trial cannot be commented upon.