This is a longitudinal, multicentre, Australia-wide observational study of patients with recurrent non-specific LBP. The study has been approved by the Human Research Ethics Committee at the Murdoch University, Perth, Western Australia (project number 2019/034).
Study flow
Eligible LBP sufferers who seek care from our participating clinicians are identified for inclusion and will be provided with an information letter and a consent form. Upon signing the consent form, the baseline questionnaire will be given to study participants to complete and be collected by clinicians. Consequently, participants will be enrolled in the SMS tracking system/software via their mobile phone numbers. Figure 1 depicts the process of recruitment and follow-up.
Data in this study will be gathered via a baseline questionnaire, frequent SMS-based questionnairs, telephone interview questionnaires, and clinicians' questionnaires.
After enrolment in the study and registration into the automated SMS system/software, each patient will receive one SMS every 2 weeks to investigate the occurrence of any new episodes of LBP. If the patient reports a new recurrent episode of pain via SMS, they will be contacted via phone by a trained member of the research team, and a telephone interview will be scheduled and performed. If a patient forgets to answer the SMS, which may happen at the beginning, in particular, a member of the research team will contact them by phone and a new explanation of the study procedure/aims will be given. This approach is necessary to ensure high compliance.
Study participants
Patients with low back pain attending primary care clinics will be encouraged to participate in this study. We intend to recruit patients aged 18 years and older. There is no predilection about the per cent of males and females participating in this study. However, it is anticipated that the proportion will be approximately equal. The sample size is 200 (see below for details) and the study is planned to be conducted Australia wide.
The following criteria are considered for inclusion in the study:
-
Patients aged 18 years and over who sought care for pain in the area between the 12th rib and buttock crease
-
History of non-specific recurrent low back pain in the past year without leg pain (participants are required to have experienced at least two episodes of LBP with the reported intensity of no less than two on the numerical rating scale (NRS) in addition to a pain-free period in between)
-
Mobile (smart) phone possession and ability to use its SMS function
-
Ability to understand, read and speak English
-
Consent, ability, and willingness to participate for the full duration of the study (52 weeks)
The following criteria will exclude a patient from participating:
-
Persistent ongoing LBP, specific LBP such as radiculopathy, disc problems, spinal canal stenosis, fractures, and metastases.
-
Secondary and tertiary care seekers
-
History of spinal surgery or severe co-morbidities such as osteoarthritis, osteoporosis, uncontrolled diabetes (to be confirmed by our participating clinicians).
-
LBP in special populations (e.g. Parkinson’s disease or pregnant women)
Participating clinicians
Different clinicians and practices (this will include chiropractors, physiotherapists, and GPs) will be approached to help recruit the study sample. Below is the list of potential organisations, practices, and clinicians, who will be contacted and asked for support:
-
Murdoch chiropractic alumni (practising anywhere in Australia)
-
Murdoch University Chiropractic Clinic (MUCC)
-
Australian Osteopathic Association
-
Chiropractic Australia
-
Australian Chiropractors Association
-
The Australian Chiropractic Research Network (ACORN)
-
Supervisors at MUCC
-
General practitioners within a 10 km radius of Murdoch University
-
Physiotherapy clinic at Curtin University of Technology
Data collection
In this study, data are being collected via different questionnaires:
Baseline questionnaire
Sociodemographic data will be collected when the patient consents to participate in the study. This consists of age, sex, height, weight, level of education, level of daily activity, an estimated number of episodes in the preceding year before the baseline, the (visual) trajectory of their pain pattern in the past year [40], and depression and anxiety state. Participants’ anxiety and depression scores will be collected via the 9-item Patient Health Questionnaire (PHQ-9) [41,42,43] and 7-item generalized anxiety disorder (GAD-7) [44, 45].
SMS questionnaire
Once enrolled, each participant receives one SMS fortnightly. Two short, simple questions will be asked in each SMS. Those questions enquire about the number of days with low back pain that had an effect on their daily activities (“troublesome”) in the past two weeks (0–14), and the severity of the pain at its worst (1–10) based on the NRS [46,47,48]. Patients will receive a reminder or a phone call if they fail to answer a SMS questionnaire.
Semi-structured telephone interview
During the telephone interview, details regarding the recent episode of low back pain will be confirmed. These include the location and duration of the recent low back pain experience. An open-ended question will be asked relating to exposure to anything during the 24 h preceding the onset of this episode that the study participants believe could have triggered the current episode. Finally, prior exposure to some well-known and previously documented triggers will be investigated through a structured interview. These are fatigue, manual tasks, moderate and vigorous physical activity, distraction during a task. The open-ended question is asked first so as not to lead the participant.
Clinicians’ questionnaire
This short questionnaire contains some basic questions regarding clinician’s name, age, sex, contact details, types of practice/registration, and years in practice (this information is collected for description purposes and determination of diversity of the participating clinicians and study population). In addition, their views on triggers of non-specific low back pain will be solicited.
Sample size calculation
As Latent Class Analysis (LCA) is not based around hypothesis testing, and there is no formal approach to calculate sample sizes. Critical factors that will affect sample size include prevalence, the sensitivity of items (i.e. class-specific prevalences), and size of classes being fitted in the model. A broadly accurate guide indicated by other researchers uses a minimum of n = 200 [49,50,51,52,53]. Three to four classes have been identified as a maximum for the outcomes being explored in this analysis. A sample size of n = 200–300 has been based on the optimal minimum sample size required for LCA, allowing for 20% contingency.
A sample size of n = 200 has 80% power to detect an odds ratio of 1.35 in adjusted logistic models comparing pain triggers against pain trajectory outcomes (intensity, frequency, length) between four groups (obtained via LCA) over 26-time points, with alpha = 0.05 and beta = 0.2. (G*Power 3.1.7).
Statistical Analysis
All descriptive summaries of patient characteristics, pain outcomes and pain triggers will include means and standard deviations or medians, interquartile ranges and ranges for continuous data and frequency distributions for categorical data. Data will be grouped according to demographic and other relevant patient factors including gender, age category, BMI category, and amount of physical activity (low, moderate, high) as well as pain pattern trajectories and triggers. All group comparisons will be performed using Chi-square tests for categorical data, and parametric one-way ANOVA or non-parametric Wilcoxon signed rank tests, depending on normality. Normality will be checked using graphical methods and Shapiro-Wilk tests. Data for patients lost to follow-up will be included up until the time of their abandonment.
Identifying pain pattern trajectories
Latent Class Analysis (LCA) will be performed in order to identify pain pattern trajectories fortnightly over 12 months based on pain intensity, frequency, and length of episodes. Summaries of patient characteristics for each pain pattern trajectory will be described and compared between trajectories.
Multinomial logistic regression or binary logistic regression using dummy variables will be used to examine patient characteristics as predictors of trajectory membership. Results will be summarised using odds ratios (ORs) and their 95% confidence intervals (CIs).
Identifying pain triggers
Pain triggers will be identified from data obtained from open- and close-ended questions in the telephone interview questionnaires. All nominated triggers will be categorised into physical and psychosocial groups.
Summaries of patient characteristics for trigger subgroups will be described and compared between trigger category subgroups such as age category, gender and physical activity categories.
Identifying triggers associated with pain pattern trajectories
Generalised linear mixed models with logit links will be used to explore the associations of individual longitudinal pain triggers with pain pattern trajectories. Significant prognostic associations between individual pain triggers and pain pattern trajectories will be identified in adjusted models that will include relevant patients’ and clinicians’ characteristics as covariates. Individual triggers that are significant at p < 0.15 in adjusted models will be entered into multivariable models and compared between trajectory groups in the final model. Results will be summarised using odds ratios (ORs) and their 95% confidence intervals (CIs).
Missing values will be accounted for in the mixed model analyses which uses maximum likelihood estimation methods. However, we require a minimum number of 12 Consecutive followups (six months) for the LCA. Stata I/C version 16.0 (StataCorp LLC, College Station TX) will be used to perform data analyses. All analyses will be 2-tailed, and p-values < 0.05 will be considered statistically significant. Sensitivity analysis will be performed, comparing patients with full follow-up data against patients with partial follow-up data.