Associations of physical activity or sedentary behaviour with pain sensitivity in young adults of the Raine Study

2.1 Participants

Cross-sectional data for this study were obtained from the Western Australian Pregnancy Cohort (Raine) Study (http://www.rainestudy.org.au). This is an ongoing birth cohort study that commenced with 2900 women who enrolled in the study before the 18th gestation week with 2,868 children born entering the initial birth cohort. Data has been collected at 1, 2, 3, 5, 8, 10, 14, 17, 20 and 22-years. The current study used data obtained at the 22-year follow up that ran between March 2012 and July 2014, 2,086 were still “active” and contacted for participation. Of these, 1,234 took part in some aspect of the 22-year follow up that included an extensive range of questionnaires and physical assessments [25], [26]. The characteristics of the active participants were compared with census data collected in 2011 on all similarly aged young adults in Western Australia and showed that the sample remains widely representative on a range of variables including education level, employment status, income, marital status, number of offspring, hours worked and occupation [25]. The ethnicity of the active participants was 85.0% Caucasian, 0.9% Aboriginal and Torres Strait Islander, and 14.1% non-Caucasian.

2.2 Recruitment, sampling and data collection

Data for this study were collected as part of 4 h of testing followed by an overnight sleep study [25]. Questionnaires were completed before physical assessments and were checked for completion by a research assistant. Anthropometry measures, and pressure and cold pain threshold testing were part of the physical assessment protocol conducted by 12 Raine research staff, all of who were thoroughly trained in the data collection procedures and used standardised protocols. For this follow up, 773 (389 female and 384 male) participants wore Actigraph GT3X+ monitors 24 h/day over a 1-week period. Participants were eligible for analysis if they had data for at least one “valid” day (≥10 h of waking wear time) and completed the Orebro Musculoskeletal Pain Questionnaire. The minimum of one “valid” day was chosen in order to maximize statistical power and to minimize selection bias. Of these, 714 individuals had valid PPT data and 702 individuals had valid CPT data.

2.3 Physical activity and sedentary behaviour

Physical activity and sedentary behaviour were objectively measured over a 1-week period using the Actigraph GT3X+ accelerometer (Actigraph, Pensacola, FL, USA) worn continuously on the right hip, except during bathing or aquatic activities. The GT3X+ was programmed to record raw data at a frequency of 30 Hz which were later reduced to vertical axis movement counts “per 60 s epoch” for the purpose of the current analyses. Accelerometer data were downloaded and processed in SAS (version 9.3, SAS Institute, Cary, NC, USA). The protocol and measured patterns of PA and SB in the Raine Study have previously been comprehensively described [6]. The “60 s epoch” was used as cut points in this age group have been validated for this length of epoch and this allows comparisons with other accelerometer data also processed using a 60 s epoch [27]. Common thresholds were used to class each minute as sedentary [<100 counts per minute (cpm)], light intensity (100–1951 cpm), moderate intensity (1952–5724 cpm) or vigorous intensity (>5,724 cpm) [6], [28]. All minutes in continuous periods of ≥90 min of zero cpm, allowing for <3 min with counts 1–50 cpm, were classed as non-wear. The algorithm used for identifying waking wear time has been reported as mostly good, and better than evaluated published alternatives [27]. Variables representing moderate PA, vigorous PA, combined moderate/vigorous PA (MVPA), sedentary time in minutes per day and sedentary time as a percentage of non-MVPA time during wake time were derived from these classifications. A further five variables captured the pattern of accumulation PA and SB as follows; MVPA min/day in ≥10 min bouts (allowing for 2 min below the cut-point), sedentary min/day in ≥20 min bouts and ≥30 min bouts, proportion of sedentary time per day accumulated in ≥20 min bouts and number of breaks from sedentary time per day.

2.4 Musculoskeletal pain status

Pain experience was determined using items from the Örebro Musculoskeletal Pain Questionnaire (ÖMPQ). The number of musculoskeletal pain areas was determined from an individual question asking “Where do you have pain?” with instruction to select appropriate sites from the options of neck, shoulder, arm, upper back, lower back, leg and other (“please state”). In the Örebro questionnaire, pain was defined as “musculoskeletal (muscle and bone) aches or pains, such as back, shoulder or neck pain”. Participants were classified by number of pain areas endorsed into “No pain areas”, “Single-site pain” and “Multisite pain” (i.e. two or more pain areas) groups. Pain chronicity was categorized from the ÖMPQ question “How long have you had your current pain problem?” into less than 3 months, 3–12 months and >12 months. Pain frequency was determined using the ÖMPQ question “How often would you say that you have experienced pain episodes, on average, during the past three months?”, using a numerical rating scale (NRS) with 1 indicating “never” and 10 indicating “always”. Pain intensity was calculated from the mean of two ÖMPQ questions “How would you rate the pain that you have had during the past week?” and “In the past three months, on average, how bad was your pain on a 0–10 scale?”, using an NRS with 1 indicating “no pain” and 10 indicating “pain as bad as it could be”.

2.5 Other variables

A number of other variables were collected at 22-years to provide a profile of participants and to control for confounders of pain sensitivity. Potential confounders of the association between pain sensitivity and PA/SB measures were considered based on a previous investigation of correlates of PPT and CPT in the Raine cohort at the 22-year follow-up [24]. Statistically significant, known, independent correlates of increased pressure pain sensitivity measures were test site, sex (female), higher waist-hip ratio (WHR) and poorer mental health [as measured by the Mental Component Summary (MCS) of the Short Form-12, version 2 (SF-12)] [24]. Statistically significant, known, independent correlates of increased cold pain sensitivity measures were sex (female), poorer mental health and smoking [24].

Waist and hip circumference were measured using a metric tape measure and standard protocol, to calculate the WHR. Health-related quality of life was measured using the SF-12 [36], a validated and reliable measure of health related quality of life. Twelve questions produce two summary measures: a MCS; and Physical Component Summary (PCS) [36]. Each SF-12 scale is a norm-based score with a mean of 50 and standard deviation of 10, with higher scores indicating better quality of life [36]. The MCS and PCS of the SF12 were categorised into those with a score ≥50 and <50. Subjects were asked, “Do you currently smoke cigarettes/cigars?” and were classified accordingly as smokers or non-smokers.

2.6 Statistical analysis

Multivariable regression models were used to examine the association between PPT and CPT measures (outcome variable) with each of the 10 PA and SB measures (independent variables). For these models, PA and SB measures were parameterised as continuous variables with the exception of moderate PA and MVPA which were categorized into deciles due to left-skew distribution, and vigorous PA and MVPA accumulated in ≥10 min bouts, which were categorized into three groups with one-third of participants registering zero activity and the remaining values split at the median of the non-zero values (1.75 and 13 min/day, respectively). The analyses for all variables were adjusted for waking wear time per day (mean of daily totals on valid days, min/day, except for “Proportion of sedentary time ≥20 min”) and number of days of valid wear time (>10 h/day). The analysis of breaks from sedentary time was adjusted for total sedentary time to reflect a break rate based on total sedentary time per day. All models were stratified by pain area groups in keeping with the aim of the study to estimate associations between PA and SB with pain sensitivity separately for people with “no pain areas”, “single-site pain” and “multisite pain”. The sample size of smallest of these groups (“single-site pain”, n=112) gave 80% power to detect increases in R² of at least 0.05 due to addition of a PA or SB variable to a base model of relevant covariates with R² of 0.15 at α=0.05. Estimates are presented with 95% confidence intervals and p-values. All models were examined for linearity of effects and absence of influential outliers, and with non-linearity modelled by addition of a quadratic term. For PPT models, linear regression models utilising generalized estimating equations with an exchangeable correlation structure to account for the repeated measures over four sites were used. PPT models were adjusted for potential confounders [24] of sex, test site, waist-hip ratio and MCS. For CPT models, Tobit regression models were used as measures were left-censored due to the lower limit of the testing equipment being 5°C. CPT models were adjusted for potential confounders [24] of sex, smoking and MCS. A sensitivity analysis was performed, to test if results of the main analysis were potentially biased by atypical activity levels as some participants with only a few valid days of accelerometer wear were included. Thus, the sensitivity analysis included only data from those participants with at least 3 valid weekdays and 1 valid day of weekend data in PPT and CPT regression models.

4.1 Strengths and limitations

Strengths of the study include sample size, age specific population, consideration of number of pain sites, control for potential correlates of pressure and cold pain sensitivity and the use of accelerometry to objectively measure PA and SB, including intensity, frequency, duration and pattern of accumulation over time [6]. The large sample at one age results in good power to estimate associations at this particular age, but the limitation is that the results may not be generalizable across age groups. Importantly, PA as measured in this study reflects habitual activity, providing a different capture of associations between pain sensitivity and PA when compared to laboratory controlled exercise protocols [14]. While previous studies using self-report measurement of PA suggest an association between higher levels of PA and decreased pressure and cold pain sensitivity [21], [22], they are limited by small participant numbers (n<72), recall bias of activity by using self-report measurement [37], and the poor correlation of self-report with objective measurement of PA [38]. Previously, only one study considered objective measurement of SB and this only included pain-free participants (n=444), finding no association between pressure and cold pain sensitivity and total daily sedentary time [24].

Affective factors potentially influence the relationship between PA and pain sensitivity, however a previous study reported that major depression did not moderate this relationship [39]. The multivariable regression models in our study were adjusted for mental health as previous investigations of the Raine cohort have reported an association of the MCS with PPT and CPT [24].

There were limitations in this study. Accelerometers were worn on the hip, therefore not measuring arm movement, were not worn while swimming and were insensitive to cycling and gradients while walking or running [40]. The authors acknowledge the limitations of an inclusion criteria of at least 1 valid day of wear time, however the sensitivity analysis including only participants with more valid days of wear time returned similar strength and direction of regression coefficients. Therefore, the inclusion criteria for wear time did not limit the results of this study.

The pressure and cold pain threshold measures used in this study may not be ideal to specifically capture the relationship of habitual PA and SB with pain sensitivity. PA can result in exercise induced hypoalgesia, with potential underlying mechanisms including acute recruitment of descending inhibitory control systems [41]. In this context, the use of dynamic quantitative sensory testing measures such as conditioned pain modulation or temporal summation may be more appropriate to capture evoked sensitivity modulation associated with PA [42]. However, conditioned pain modulation and exercise induced hypoalgesia have been found to be partially impaired in chronic pain patients with high versus low pressure pain sensitivity [16].

The literature suggests the number of pain sites is an important factor to consider when investigating the relationship between PA and pain sensitivity [15], [17], hence in the current study, participants were categorized according to their current pain status, so chronicity of pain was not considered, meaning the “Single-site pain” and “Multisite pain” groups contained participants with pain of varying duration. Table 1 reports the “Multisite pain” group contained participants with higher levels of pain chronicity, pain frequency and pain intensity when compared to the “Single-site pain” group.

Numerous statistical contrasts were performed without adjustment of the type I error rate, adopting the philosophy of Sterne et al. [43] of the unadjusted p-value as strength of evidence against the null hypothesis, and the 95% confidence interval as the range of credible values for the population parameter. It is possible that the few associations observed in this study occur by chance only, and the confidence intervals for these estimates indicate that differences may not be of a meaningful magnitude. Furthermore, the associations detected are only cross-sectional, and give us no information as to how PA and SB behaviours might temporally heighten or lower pressure and cold pain sensitivity. The following discussion of the associations identified by this study is therefore presented with this caveat in mind.

4.2 Pain sensitivity, physical activity and sedentary behaviour

With respect to findings regarding pressure pain sensitivity, for the “Single-site pain” group, participants with higher levels of MVPA accumulated in ≥10 min bouts (>13 min/day) demonstrated greater pressure pain sensitivity compared with those participants with ≤13 min/day MVPA accumulated in ≥10 min bouts, but not compared to those with no MVPA accumulated in ≥10 min bouts. These findings suggest that for participants with “single site pain”, how MVPA is accumulated (min accumulated in longer bouts of MVPA) may be important in the context of heightened pressure pain sensitivity. The mechanisms underlying heightened pressure pain sensitivity and PA in the “single site pain” group are likely complex, potentially involving both neuronal [44] and non-neuronal factors (e.g. immune) [45]. Given pressure pain sensitivity was measured across four sites and models were adjusted for site, this association might plausibly reflect changes in central nociceptive processing or modulation (for example, altered endogenous descending control system efficiency) [14] or facilitated spatial/temporal summation in response to PA [16], rather than primarily peripheral sensitisation (as this would manifest in a more localised site sensitivity). However, it is unclear why this association would be detected for the single-site pain group, but not multisite pain. Variable effects of PA on pressure pain sensitivity in both clinical and experimental pain populations have been reported [15], but interpretation is complicated by differences in study quality, design, exercise protocols, measurement tools, clinical populations and outcomes [14].

With respect to findings on cold pain sensitivity, for those with “multisite pain”, participants falling within the highest tertile of vigorous PA (VPA) had greater cold pain sensitivity when compared with participants with lower or no vigorous PA. It is unclear what this association might reflect, as in this “multisite pain group”, similar differences between VPA levels for pressure sensitivity would also be expected, given the potential for facilitated (temporal and spatial) nociception from deep tissues following exercise in multisite pain (for example in chronic widespread pain, or fibromyalgia [16], [17]. Notwithstanding this point, differences in cold sensitivity levels have been demonstrated previously in a non-clinical cohort drawn from the Raine Study (young females), with heightened cold pain sensitivity evident in those females reporting moderate to severe menstrual pain [46] and an association between low cortisol response to stress and musculoskeletal pain in females with heightened cold pain sensitivity [13]. These authors suggest that cold hypersensitivity may reflect changes in central regulatory systems linked to homeostasis (including thermosensation and thermoregulation). It is also possible that VPA in this group may differentially influence cold and pressure pain sensitivity, as these psychophysical tests are designed for nociceptors located in skin and muscle tissue, respectively [32]. Collectively, these findings allude to potentially important dose-relationships between PA/exercise and pain sensitivity, suggesting that higher amounts of VPA may not be ideal for all musculoskeletal pain conditions, particularly for clinical populations with two or more pain areas [15], [17].

The association of lower cold pain sensitivity with an increase in the number of breaks from sedentary time for participants in the “No pain areas” group also suggests the way sedentary time is accumulated may be related to pain sensitivity. Increased breaks in sedentary time, independent of total sedentary time, have demonstrated associations with lower waist circumference [47], [48], lower inflammatory marker concentration [47] and improved plasma glucose levels [47], [48]. These physiological effects may suggest mechanisms whereby more breaks from sedentary time could be associated with lower cold pain sensitivity partly through mechanisms including improved energy metabolism and lower circulating inflammatory markers. Young adults spend most of the waking day being sedentary [6] and targeted interventions for pain prevention and also for improving other life-course health trajectories, should consider the accumulation patterns of sedentary time.