Study design

We hypothesised that participants with higher levels of physical activity (considering duration and intensity of activity) before the index episode would be more likely to recover within the first 12 months. We also hypothesized that those who recovered from their pain episode would be more likely to return to their original (i.e. pre pain) levels of physical activity participation. We used longitudinal data from the TRIGGERS for low back pain Study. The study was approved by the Human Research Ethics Committee of the University of Sydney (protocol number 05-2011/13742). The TRIGGERS Study was a case-crossover study designed to investigate a number of transient physical and psychosocial risk factors for an episode of sudden-onset, acute low back pain. Follow-up assessment was completed at 3 and 12 months following the index episode of pain, through telephone interviews. All participants were assessed by the same team of researchers, following the same standardised assessment booklet. Detailed information has been published elsewhere [13].

Participants

The study included 999 consecutive patients, aged 18 years or older, that presented to 300 primary care clinics (general practitioners, physical therapists, and chiropractors) with a new episode of acute low back pain (i.e. index episode). All participants gave written informed consent for participation. Participants were referred by the clinics and recruited between October 2011 and November 2012 and interviewed to assess the inclusion/exclusion criteria previously described in the study protocol [14]. A new episode of acute low back pain was defined as a primary complaint of pain between the 12th rib and the buttock crease, with or without leg pain, causing the patient to seek health care, and preceded by a period of at least 1 month without back pain [15]. Patients needed to present to primary care within 7 days from index episode and report pain of at least moderate intensity in the first 24 hours (measured using item 7 of the Short Form 36 questionnaire: “How much bodily pain have you had during the past 4 weeks? None; Very mild; Mild; Moderate; Severe or Very severe.”) [16]. Participants with known or suspected serious spinal pathology (metastatic, inflammatory or infective diseases of the spine) were excluded.

Outcome measures

At study entry, a telephone-based interviewed was conducted within 7 days following referral to the study. At baseline, participants were asked about their age, gender, height, weight, medication taken for the pain, employment status, type of work, manual tasks performed (e.g. lifting heavy loads, awkward positioning, handling objects far from the body), a recent slit, trip or fall, pain intensity, feelings of depression or anxiety, alcohol consumption, as well as engagement in moderate or vigorous physical activity and sexual activity. The date and time of pain onset was also identified with the assistance of recommended recall aids, such as a diary and/or smartphone. Participants’ BMI was calculated based on the self-reported height and weight. Participants were also asked to report pain intensity at the moment of the interview using a numeric rating scale (0 to 6) and disability (0 to 5) based on item 7 of the SF-36 questionnaire [17]. The SF-36 is a valid and reliable self-administered tool for measurement of physical and social functioning [17]. The SF-36 correlation coefficient with the International Quality of Life Assessment (IQOLA) for the physical functioning section is 0.85 and 0.42 for the social functioning section, and the internal consistency reliability coefficient for the physical functioning is 0.93 and 0.68 for social functioning.[18]. The overall internal consistency reliability of SF-36 is estimated to be 0.87 [19]. Tension or anxiety and feelings of depression were assessed based on items 13 and 14 of the Orebro Musculoskeletal Pain Questionnaire (“How tense or anxious have you felt in the past week?” and “How much have you been bothered by feeling depressed in the past week?”, respectively) [20].

The self-reported level of physical activity participation was assessed using the Active Australia Survey (AAS) questionnaire [21], where participants estimated the total number of hours they spent engaged in light, moderate, and vigorous physical activity in the week preceding the index episode, and at 1 week and at 12 months follow-up. Participants were contacted again at 3 months and 12 months and were asked if they had recovered from the index episode, and in case of a positive answer, the date when the pain fully resolved was used to calculate the duration of episode. Recovery was defined as being pain-free for at least one month before the follow-up interview [15]. This design has already been successfully used to assess physical activity participation among persons seeking care due to low back pain [22]. Participants were also asked at baseline and follow-up surveys about the presence and intensity of symptoms of depression and anxiety.

Assessment of frequency and intensity of physical activity

In this study, physical activity was defined as any bodily movement that increases the metabolic demand above the rest state [23]. Participants were asked about the total time spent in physical activity of different intensities (e.g. light, moderate or vigorous) the week before pain onset. Pain onset could have been up to 2 weeks prior to the baseline interview. The AAS questionnaire was developed focusing on the recall period of 1 week and has been shown to present estimates of physical activity level comparable with pedometer steps and accelerometer counts and considers physical activity of light intensity (e.g. walking briskly), moderate intensity (e.g. gentle swimming and recreational tennis) and vigorous intensity (e.g. cycling and competitive tennis) [24]. The AAS was developed based on the International Physical Activity Questionnaire (IPAQ) therefore we used the IPAQ-SF scoring system to calculate the metabolic demand and total metabolic expenditure minutes (MET min/week) following the scoring system used in the short form of the IPAQ (IPAQ-SF) for different levels of intensity of physical activity (i.e. light = 3.3 met/min, moderate = 4 met/min and vigorous = 8 met/min) [25] for each participant at each time point.

Statistical analysis

Patients were split into two groups according to the presence of pain at the 12-month follow-up assessment; ‘recovered’ if they no longer reported pain and ‘not recovered’ if they reported persistent pain. Baseline demographic variables were summarised as mean (SD) if continuous and frequency (percent) if categorical.

Generalised estimating equations were used to investigate the change in total metabolic equivalent minutes over time. A time-by-group interaction term was included alongside main effects recovery group and time-point (one-week prior to episode, week of episode and 12 months after episode) were to conduct within- and between-group comparisons at each time point as well as across all time points.

Logistic regression models were used to assess the association between levels of physical activity one-week prior to the pain episode and the risk of non-recovery at 12 months.

All models were adjusted for known confounders including age, gender, BMI, pain intensity at index episode and depression or anxiety symptoms assessed based on Orebro Musculoskeletal Pain Questionnaire [8, 20].

Total Metabolic Equivalent minutes per week (MET min/week) was used to quantify physical activity participation prior to index episode and used in the analysis to predict the presence of low back pain at the 12-month follow-up. We used the scoring system from the IPAQ-SF since it has been validated against accelerometer data [25] and it is a well-known way to pool physical activities of different intensities to facilitate the comparison between different activities [26]. Generalised linear models were conducted using total MET min/week as the dependent variable (continuous) considering each time-point (1 week before low back pain episode, week of the low back pain episode and 12 months after the episode) and the presence of low back pain at 3 months and 12 months.

A sensitivity analysis was conducted, repeating the analyses on the 3-month follow-up assessment [27]. All analyses were conducted using IBM SPSS Statistics for Windows, Version 22.0 (Armonk, NY: IBM Corp.) Statistical significance was set as p < 0.05.

Change in physical activity participation

For the physical activity participation analyses, participants were grouped according to their pain status at 12 months. Both groups (i.e. participants with persistent pain and participants who had recovered) reported similar levels of physical activity participation at all time-points assessed: one week before the index episode (p = 0.449), one week after index episode (p = 0.812) and 12 months after the index episode (p = 0.233) (Table 2). Participants reported a statistically significant reduction in the mean MET minutes/week one week after the index episode in both persistent pain (mean METs minutes/week: -788; 95% confidence interval [CI]: -990 to -585; p<0.001) and recovered (METs minutes/week: -611; 95% CI: -723 to -498; p<0.001) groups. At 12 months, both groups reported increases in the levels of physical activity participation similar to pre-pain levels (Fig 2). The generalised linear model confirmed a similar pattern for physical activity participation between groups after adjusting for age, gender, BMI, pain intensity at event and depression, at both 3 and 12 months after the index episode (χ² = 3.622, p = 0.16 and χ² = 2.963, p = 0.23, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mean total MET minutes per week at each time-point.

p value > 0.05 for all comparisons.

https://doi.org/10.1371/journal.pone.0219556.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Difference in physical activity level (total MET minutes per week) before the onset of low back pain within group among multiple time-points.

https://doi.org/10.1371/journal.pone.0219556.t002

The results of our logistic regression models showed that baseline physical activity participation (i.e., before index episode) was not a predictor of recovery at 12 months, after adjusting for age, gender, BMI, pain intensity at baseline and depression or anxiety symptoms (AOR 0.99; 95% CI 0.99 to 1.00; p = 0.523)(Table 3). Similarly, when considering those who had recovered within the first 3 months post index episode, our sensitivity analysis showed that the amount of physical activity before index episode did not predict recovery at 3 months (OR 1; 95% CI 0.99 to 1.00; p = 0.923) (Table 3). The results of unadjusted logistic regression models are presented in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between physical activity level (total MET minutes per week) before index episode and the presence of low back pain 3 months and 12 months after the index episode.

https://doi.org/10.1371/journal.pone.0219556.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Association between physical activity level (total MET minutes per week) before index episode and the presence of low back pain 3 months and 12 months after the index episode (unadjusted analysis).

https://doi.org/10.1371/journal.pone.0219556.t004

Study limitations

Our study presents some limitations that need to be considered when interpreting the results. For instance, 17% of participants were lost to follow-up due to unavailability to respond or declining to participate in the study. We also acknowledge that physical activity participation was ascertained via questionnaire. There is evidence suggesting that self-reported measures of physical activity participation result in overestimation of participation when compared to accelerometer-based data, especially in cohorts with moderate to high intensity physical active routines [32]. The Active Australia questionnaire also relies on recall, as participants were asked about their level of physical activity on the week before the interview; however, short recall periods of up to 2 weeks are reported as less susceptible to recall bias [21]. This could be a potential source of bias since the perception of pain could lead the participants to change their perceptions of how much physical activity they performed on the week prior to the pain, however, despite the maximum recall time possible in our study being of up to 3 weeks, the actual mean recall time was of only 5 days [13]. Due to the high variability of the physical activity measurement and the recall bias, these could have biased the results of the assessed associations towards the null.