Advertisement

Prognostic Indicators of Low Back Pain in Primary Care: Five-Year Prospective Study

Open AccessPublished:June 21, 2013DOI:https://doi.org/10.1016/j.jpain.2013.03.013

      Abstract

      Back pain is common and many people experience long-term problems, yet little is known about what prognostic factors predict long-term outcomes. This study's objective was to determine which factors predict short- and long-term outcomes in primary care consulters with low back pain (LBP). Analysis was carried out on 488 patients who had consulted their physician about LBP. Patients were followed up at 6 months and 5 years. Clinically significant LBP at follow-up was defined as a score of 2, 3, or 4 on the Chronic Pain Grade, indicating substantial pain and disability. Cox regression was used to estimate relative risks (RRs) with 95% confidence intervals (CIs) on 32 potential predictive factors, organized into domains (demographic, physical, psychological, and occupational). Baseline pain intensity conferred a 12% increase in risk (RR = 1.12, 95% CI = 1.03–1.20), and patients' belief that their LBP would persist conferred a 4% increase in risk (RR = 1.04, 95% CI = 1.01–1.07) for poor outcome at 6 months. Outcome at 5 years was best predicted by a model with the same factors as in the 6-month model: pain intensity increased risk by 9% (RR = 1.09, 95% CI = .997–1.20), and a belief that their LBP would persist increased risk by 6% (RR = 1.06, 95% CI = 1.03–1.09). Both predictors have the potential to be targets for clinical intervention.

      Perspective

      Few studies have investigated factors that predict long-term back pain. This study has shown that pain intensity experienced during a period of primary care consultation, and patients' perception about whether their back pain will persist, were significant predictors of poor outcome at 6 months and at 5 years.

      Key words

      The prevalence of low back pain (LBP) is substantial, with population estimates of 50 to 70% over a lifetime.
      • Leboeuf-Yde C.
      • Nielsen J.
      • Kyvik K.O.
      • Fejer R.
      • Hartvigsen J.
      Pain in the lumbar, thoracic or cervical regions: Do age and gender matter? A population-based study of 34,902 Danish twins 20-71 years of age.
      • Papageorgiou A.C.
      • Croft P.R.
      • Ferry S.
      • Jayson M.I.
      • Silman A.J.
      Estimating the prevalence of low back pain in the general population. Evidence from the South Manchester Back Pain Survey.
      Up to half the people with LBP seek health care for their pain.
      • Leboeuf-Yde C.
      • Nielsen J.
      • Kyvik K.O.
      • Fejer R.
      • Hartvigsen J.
      Pain in the lumbar, thoracic or cervical regions: Do age and gender matter? A population-based study of 34,902 Danish twins 20-71 years of age.
      Evidence highlights that many people with LBP do not have single episodes but often experience long-term pain with significant recurrence and fluctuations.
      • Hayden J.A.
      • Dunn K.M.
      • van der Windt D.A.
      • Shaw W.S.
      What is the prognosis of back pain?.
      • Hestbaek L.
      • Leboeuf-Yde C.
      • Manniche C.
      Low back pain: What is the long term course? A review of studies of general patient populations.
      • Leboeuf-Yde C.
      • Nielsen J.
      • Kyvik K.O.
      • Fejer R.
      • Hartvigsen J.
      Pain in the lumbar, thoracic or cervical regions: Do age and gender matter? A population-based study of 34,902 Danish twins 20-71 years of age.
      • Picavet H.S.J.
      • Schouten J.S.
      Musculoskeletal pain in the Netherlands: Prevalences, consequences and risk groups, the DMC3-study.
      This results in considerable costs for health care and society.
      • Manchikanti L.
      • Singh V.
      • Datta S.
      • Cohen S.P.
      • Hirsch J.A.
      Comprehensive review of epidemiology, scope, and impact of spinal pain.
      One important area of focus within LBP research is the identification of key prognostic factors.
      • Hayden J.A.
      • Dunn K.M.
      • van der Windt D.A.
      • Shaw W.S.
      What is the prognosis of back pain?.
      There is a diverse range of prognostic factors in relation to LBP: demographics such as educational status, age, and gender,
      • Hoy D.
      • Brooks P.
      • Blyth F.
      • Buchbinder R.
      The epidemiology of low back pain.
      physical factors such as the level of pain intensity and disability perceived by the patient,
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      psychological factors such as depression and anxiety
      • Main C.J.
      • Foster N.
      • Buchbinder R.
      How important are back pain beliefs and expectations for satisfactory recovery from back pain.
      and pain-specific concepts such as fear avoidance, catastrophizing, and illness perceptions,
      • Burton A.K.
      • McClune T.D.
      • Clarke R.D.
      • Main C.J.
      Long-term follow-up of patients with low back pain attending for manipulative care: Outcomes and predictors.
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      and occupational factors such as employment status.
      • Dunn K.M.
      • Jordan K.P.
      • Croft P.R.
      Contributions of prognostic factors for poor outcome in primary care low back pain patients.
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      Importantly, these factors can characterize groups of people at higher risk of persistent pain and disability, and they highlight potentially modifiable factors to target in clinical interventions (eg, psychological therapies and occupational interventions).
      • Hayden J.A.
      • Dunn K.M.
      • van der Windt D.A.
      • Shaw W.S.
      What is the prognosis of back pain?.
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      However, most prognostic studies of LBP have considered follow-up periods of 1 year or less (see reviews
      • Hestbaek L.
      • Leboeuf-Yde C.
      • Manniche C.
      Low back pain: What is the long term course? A review of studies of general patient populations.
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      ). For example, of the 32 studies on back or spinal pain included in a review by Mallen and colleagues,
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      only 3 had follow-up periods longer than 1 year. This is problematic, as potential prognostic factors could differ depending on the time scale.
      • Burton A.K.
      • McClune T.D.
      • Clarke R.D.
      • Main C.J.
      Long-term follow-up of patients with low back pain attending for manipulative care: Outcomes and predictors.
      • Dunn K.M.
      • Jordan K.
      • Croft P.R.
      Characterizing the course of low back pain: A latent class analysis.
      • Hayden J.A.
      • Dunn K.M.
      • van der Windt D.A.
      • Shaw W.S.
      What is the prognosis of back pain?.
      For example, one study (Burton et al
      • Burton A.K.
      • McClune T.D.
      • Clarke R.D.
      • Main C.J.
      Long-term follow-up of patients with low back pain attending for manipulative care: Outcomes and predictors.
      ) followed up patients attending private group osteopathic practices. They tested factors that were associated with disability at 1 year and at 4 years and report that fear avoidance, passive coping, and catastrophizing were significant at 1 year, but depression and pain intensity were significant at 4 years. They suggested that initial fear avoidance, catastrophizing, and passive coping possibly give way to depression in the long term. However, another recent study
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      considered differences in prognostic factors between primary care patients with acute/subacute pain (defined as pain duration of less than 3 months prior to baseline assessment) and those with chronic pain (defined as pain duration of more than 3 months prior to baseline assessment) at baseline. They reported no differences in prognostic factors between these groups in the prediction of disability 12 months later. This clearly shows that further study is required to understand the potential for differences in prognostic markers dependent on time. Indeed possible differences in prognostic factors over time may be a reason why current interventions for LBP show low sustainability of treatment effect over the long term.
      • Artus M.
      • van der Windt D.A.
      • Jordan K.P.
      • Hay E.M.
      Low back pain symptoms show a similar pattern of improvement following a wide range of primary care treatments: A systematic review of randomized clinical trials.
      We need to better characterize factors that independently predict short- and long-term outcome of patients with LBP in order to inform and test treatments that target different prognostic groups.
      The aim of this study is to investigate, in patients with LBP consulting in a primary care setting, which prognostic factors predict poor pain and disability outcomes 5 years later and to compare these with predictors of earlier short-term outcomes at 6-month follow-up in the same cohort.

      Methods

      Design and Setting

      Participants in a large prospective cohort study of persons visiting their primary care physician about LBP were mailed questionnaires soon after their healthcare visit (baseline) and again 6 months and 5 years later. The population for this analysis were responders to the baseline questionnaire (N = 1591) who gave consent to further contact and who responded again at 6 months (n = 810) and 5 years (n = 488).
      Ethical approval was given by North Staffordshire and North West Cheshire Research Ethics Committees for all phases of the study.

      Recruitment and Procedure

      Patients, aged between 18 and 60 years, who visited their primary care physician about LBP at 8 primary care practices within the North Staffordshire and Cheshire area of England were invited to take part.
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      Primary care practices are the gateway to the healthcare system within the United Kingdom. The practices cover a range of deprivation areas, and given that more than 95% of the UK population is registered with a primary care practice,
      • Bowling A.
      Research Methods in Health.
      they are representative of the local population. At baseline, eligible participants were identified via computerized primary care records using the “Read Code” system, which is the standard method of coding and recording reasons for contact in UK general practice. All codes relating to LBP were used to identify potential participants, with specific codes for “red flag” diagnoses (cauda equina syndrome, significant trauma, ankylosing spondylitis, cancers) used as exclusion criteria. Quality and validity of the Read Code system within these practices is assessed annually through continual training and feedback to ensure high levels of recording of relevant Read codes during patient healthcare visits.
      • Porcheret M.
      • Hughes R.
      • Evans D.
      • Jordan K.
      • Whitehurst T.
      • Ogden H.
      • Croft P.
      North Staffordshire general practice research network. Data quality of general practice electronic health records: The impact of a program of assessments, feedback, and training.
      The target cohort consisted of 1,591 adults who had visited their primary care physician for LBP in the study practices and who responded to the initial baseline questionnaire mailed to them within 2 weeks of their index visit. We have previously shown that this cohort is broadly representative of all patients attending primary care for LBP in these practices
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      and that the registered populations are broadly representative of a UK population generally. Of the 1,591 back pain patients recruited at baseline, 810 completed and returned their 6-month follow-up questionnaire and 488 responded at 5 years (70% of those eligible). This cohort of 488 responders at 5 years formed the basis for the analyses presented in this paper (see flow diagram Fig 1).

      Measures

      Outcome Measure

      Pain and disability related to LBP were measured at 6 months and at 5 years using the Chronic Pain Grade (CPG).
      • Von K.M.
      • Miglioretti D.L.
      A prognostic approach to defining chronic pain.
      • Von K.M.
      • Ormel J.
      • Keefe F.J.
      • Dworkin S.F.
      Grading the severity of chronic pain.
      The CPG is a 7-item measure of chronic pain for assessing the patient using 2 dimensions: pain severity (pain at present time and worst and average pain over the previous 6 months) and disability (pain interference, social and employment activity restriction due to pain over the previous 6 months). The measure categorizes outcome of chronic pain and disability into 5 grades: Grade 0 (pain free), Grade 1 (low disability, low pain intensity), Grade 2 (low disability, high pain intensity), Grade 3 (high disability, pain that is moderately limiting), and Grade 4 (high disability, pain that is highly limiting). The CPG has established validity in the UK population
      • Smith B.H.
      • Penny K.I.
      • Purves A.M.
      • Munro C.
      • Wilson B.
      • Grimshaw J.
      • Chambers A.W.
      • Cairns Smith W.
      The Chronic Pain Grade questionnaire: Validation and reliability in postal research.
      and as a measure of pain over time.
      • Elliott A.M.
      • Smith B.H.
      • Smith W.C.
      • Chambers W.A.
      Changes in chronic pain severity over time: The Chronic Pain Grade as a valid measure.
      For the purpose of this study, CPG scores were collapsed to form 2 groups, with Grades 0 and 1 forming a group with no or low levels of pain or disability and Grades 2, 3, and 4 forming a clinically significant LBP group, following previous methodology,
      • George C.
      The six-month incidence of clinically significant low back pain in the Saskatchewan adult population.
      • Von K.M.
      • Miglioretti D.L.
      A prognostic approach to defining chronic pain.
      with the latter being defined as a poor outcome.

      Demographic Predictors

      Participants were grouped into 1 of 4 age categories—≤37, 38 to 45, 46 to 52, and ≥53 years—because of the nonlinear relationship of age and LBP
      • Hoy D.
      • Brooks P.
      • Blyth F.
      • Buchbinder R.
      The epidemiology of low back pain.
      ; gender was used as a dichotomized variable. Educational status was categorized into 2 groups, education up to the age of 16 and educated beyond the age of 16, and social class was dichotomized
      • Office for National Statistics
      Standard occupational classification. The Codingindex, vol. 2, The Stationary Office, Version 1.1, The National Statistics Socio-economic Classification User Manual.
      into higher (managerial, professional, intermediate, self-employed) and lower (supervisory, technical, routine occupations), as used previously.
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.

      Physical Predictors at Baseline

      LBP disability was assessed using the 24-item Roland Morris Disability Questionnaire (RMDQ).
      • Roland M.
      • Morris R.
      A study of the natural history of back pain. Part I: Development of a reliable and sensitive measure of disability in low-back pain.
      The RMDQ asks questions on the level of disability associated with LBP on the day of questioning and gives a score from 0 to 24 (a higher score indicates a higher level of disability).
      Pain intensity was measured by calculating the mean of 3 numerical rating scales (0–10) for the participant's least and usual LBP over the previous 2 weeks, and their rating of current pain intensity at the time of filling in the questionnaire.
      • Dunn K.M.
      • Jordan K.
      • Croft P.R.
      Characterizing the course of low back pain: A latent class analysis.
      A higher score indicates a higher level of reported pain intensity.
      Symptom duration has been shown to be an important prognostic marker for poor outcome.
      • Dunn K.M.
      • Jordan K.P.
      • Croft P.R.
      Recall of medication use, self-care activities and pain intensity: A comparison of daily diaries and self-report questionnaires among low back pain patients.
      Pain duration at baseline was measured by asking participants to recall when they had their last “pain-free” month. Two groups were created: 1) pain duration less than 3 years prior to baseline and 2) pain duration more than 3 years prior to baseline. This choice of timeline (duration of 3 years) was based on a recent prognostic study that included the influence of prior to baseline indication of pain duration (ie, how long patient had pain before entering the study) as a risk factor. They reported no differences in outcome (in their case recovery) using the usual 3 months acute-chronic distinction; however, they did report differences on outcome using the above- or below-3-year category.
      • Dunn K.M.
      • Croft P.R.
      The importance of symptom duration in determining prognosis.
      The presence of leg pain and distal pain (eg, above, below knee) and of upper body pain (shoulder, arm, neck, or head) was also recorded.
      • Dunn K.M.
      • Croft P.R.
      The importance of symptom duration in determining prognosis.
      • Dunn K.M.
      • Jordan K.P.
      • Croft P.R.
      Recall of medication use, self-care activities and pain intensity: A comparison of daily diaries and self-report questionnaires among low back pain patients.

      Psychological Predictors

      The Illness Perception Questionnaire–Revised (IPQ-R) was used to assess illness perceptions.
      • Moss-Morris R.W.
      • Weinman J.
      • Petrie K.J.
      • Horne R.
      • Cameron L.D.
      • Buick D.
      The revised Illness Perception Questionnaire (IPQ-R).
      The IPQ-R has 12 subscales, 8 for illness perceptions (illness identity, consequences, timeline–acute to chronic, timeline–cyclical, illness coherence, treatment control, personal control, emotional representations) and 4 on the causes of LBP (accident/chance, psychological, risk factors, immunity). Higher scores on each subscale of the IPQ-R indicate stronger illness perceptions, with some intersubscale items being reverse scored.
      Fear of movement was measured by the Tampa Scale for Kinesiophobia (TSK).
      • Kori S.H.
      • Miller R.P.
      • Todd D.D.
      Kinesiophobia: A new view of chronic pain behaviour.
      The TSK contains 17 items about a person's fear of movement because of pain, with higher scores indicating a higher level of movement avoidance.
      The Coping Strategies Questionnaire 24 (CSQ-24) was used to assess coping styles in participants.
      • Harland N.J.
      • Georgieff K.
      Development of the Coping Strategies Questionnaire 24, a clinically utilitarian version of the Coping Strategies Questionnaire.
      Four scales are used in the questionnaire (catastrophizing, diversion, reinterpretation, cognitive-coping), with higher scores indicating a higher frequency of use of the coping style.
      In addition, the baseline questionnaire contained 12 items on behavioral coping in relation to the patient's pain (use of over-the-counter medicine, lying down, creams or sprays, hot and cold packs, massage, lumbar support, walking, swimming, general practitioner-prescribed medication, walking stick, bed rest, exercise). Using guidance from Brown and Nicassio's active and passive coping model,
      • Brown G.K.
      • Nicassio P.M.
      Development of a questionnaire for the assessment of active and passive coping strategies in chronic pain patients.
      exploratory factor analysis was performed on these items. Analysis revealed suitability of these items to factor analysis, and a 2-factor “active” and “passive” coping construct was created.
      • Zadurian N.
      The role of coping styles and strategies in patients consulting for low back pain in primary care.
      The Pain Self-Efficacy Questionnaire (PSEQ) was used to measure participants' beliefs and confidence in their ability to accomplish activities and engage in situations (eg, doing household chores, being active, getting enjoyment out of things, and leading a normal life) despite their level of pain.
      • Nicholas M.K.
      The pain self-efficacy questionnaire: Taking pain into account.
      The measure consists of 10 items, each scored by a 6-point Likert-type scale, with a higher score indicating greater self-efficacy.
      The Hospital Anxiety and Depression Scale (HADS) was used to measure depressive and anxiety symptoms.
      • Zigmond A.S.
      • Snaith R.P.
      The Hospital Anxiety and Depression Scale.
      The measure consists of 14 questions and produces a scale from 0 to 21 for depression and anxiety separately, and is applicable to general population samples.

      Occupational Predictors

      Four groups were created on the basis of the reports of the respondents current work status: 1) currently employed in normal job, 2) currently employed on reduced duties because of LBP, 3) currently unemployed because of LBP, and 4) currently unemployed because of other reasons.

      Statistical Analysis

      Univariate regression was performed on all predictors in relation to poor outcome at 6 month and at 5 years. Cox regression, with a constant time variable,
      • Thompson M.L.
      • Myers J.E.
      • Kriebel D.
      Prevalence odds ratio or prevalence ratio in the analysis of cross sectional data: What is to be done?.
      was used to determine an unadjusted regression coefficient (relative risk [RR]) for each predictor.

      Lumley T, Kronmal R, Shuangge M: Relative risk regression in medical research: Models, contrasts, estimators and algorithms. UW Biostatistics Working Paper Series, paper 293. Available at: http://www.bepress.com/uwbiostat/paper293. Accessed February 2013

      All predictors with P values < .1 at this stage were retained for multivariable analysis. Prognostic factors were then examined within 4 domains (ie, demographic factors, physical factors, psychological factors, and occupational factors). Domains were chosen to coherently manage the large number of variables within the data set, and to extract the predictor or predictors that best explain the association with poor outcome from the respective domains (ie, the key demographic factors, physical factors, and psychological factors). For example, Pincus et al
      • Pincus T.
      • Burton A.K.
      • Vogel S.
      • Field A.P.
      A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain.
      has suggested that many psychological variables related to pain overlap conceptually. Therefore, this current approach allowed for mutual adjustment to account for statistical overlap that may be present between variables (intercorrelation) and to identify the significant variable or variables for that domain. All nonsignificant predictors (P > .05) in each domain-specific model were removed. The remaining predictors from each domain-specific model were then entered into a final multivariable model with each other for outcomes at 6-month and 5-year follow-up time points, which produced adjusted RRs with 95% confidence intervals (CIs).
      Missing data and the impact of nonresponse at baseline and follow-up was analyzed by comparing participants' baseline response on individual prognostic factors with response status at 6 months and 5 years, using summary measures (Appendix 1). Checks for multicollinearity were carried out because of the large number of variables with the potential to have statistical overlap, which can then affect the unique estimates of each variable and lead to instability within the coefficient estimates. Following guidelines for multivariable analysis, each variable was set sequentially as a dependent variable, and all others were set as independent variables using linear regression.
      • Katz M.H.
      Multivariable Analysis: A Practical Guide for Clinicians.
      Variance inflation factor (VIF) values were inspected to indicate the regression coefficient influence from all other variables; variables with VIF scores >4 would be considered problematic.

      Results

      Baseline characteristics show a mean age of the cohort of 47.4 years, with just over 62% being female. More than half (58.6%) indicated being in education beyond 16 years of age and just over 60% were employed. More than 22% describe experiencing their LBP for more than 3 years, with over 60% describing pain spreading to their legs. Full details for the study participants included in this analysis are presented in Table 1. No variables were found to be highly related (ie, VIF score >4).
      Table 1Baseline Characteristics of 488 LBP Patients Seeking Consultation
      Baseline CharacteristicsNumber (%)Mean (SD)Interquartile Range
      Demographic
       Age (years)47.4 (9.0)41–45
       Gender (female)303 (62.1)
       Education (beyond 16)286 (58.6)
       Social class (low)184 (37.7)
      Physical
       Pain intensity3.9 (2.3)2.3–5.3
       RMDQ8.5 (5.9)4–13
       Pain duration (>3 years)109 (22.3)
       Leg pain (yes)299 (61.3)
       Distal pain (yes)168 (34.4)
      Psychological
       HADS, Anxiety symptoms8.1 (4.4)5–11
       HADS, Depressive symptoms6.2 (4.2)3–9
       TSK, Fear of movement39.0 (6.9)35–43
       CSQ, Catastrophizing9.2 (7.4)3–13
       CSQ, Diversion16.0 (8.2)10–21
       CSQ, Reinterpretation7.5 (6.8)2–11
       CSQ, Cognitive coping17.0 (6.0)13–21
       IPQR, Symptoms4.1 (2.3)2–5
       IPQR, Consequences17.2 (5.5)13–21
       IPQR, Cyclical time13.0 (3.3)11–16
       IPQR, Emotional representation16.5 (5.2)13–20
       IPQR, Illness coherence13.1 (5.0)10–17
       IPQR, Personal control21.1 (3.7)18–24
       IPQR, Treatment control17.2 (3.2)15–20
       IPQR, Timeline acute-chronic19.9 (5.7)16–24
       IPQR, Psychological attributions11.6 (3.9)9–14
       IPQR, Risk factor attributions15.0 (4.0)12–17
       IPQR, Immunity dimension5.2 (1.9)3–6
       IPQR, Accident chance5.9 (1.9)5–7
       Pain self-efficacy39.0 (6.9)29–50
       Active behavioral coping1.2 (.9)0–2
       Passive behavioral coping2.2 (1.4)1–3
      Occupational
       Employed294 (60.9)
       Employed, reduced duties due to LBP74 (15.3)
       Unemployed78 (16.1)
       Unemployed due to LBP37 (7.7)
      At 6 months, 47.7% had clinically significant LBP, falling to 36.9% at the 5-year follow-up. The regression models showing the baseline factors predicting clinically significant LBP status at the 6-month and 5-year follow-ups are presented in Table 2.
      Table 2Cox Regression Models for the Relationship Between Prognostic Indicators at Baseline and Clinically Significant LBP Group at 6-Month and 5-Year Follow-Up Stages
      Prognostic IndicatorsUnadjusted RR (95% CI)Domain Adjustment RR (95% CI)Final Model RR (95% CI)
      6-Month5-Year6-Month5-Year6-Month5-Year
      Demographics
       Age, years (<38 reference)
      38–45.949 (.64, 1.4).959 (.60, 1.5)
      46–52.949 (.67, 1.4).933 (.59, 1.5)
      >521.05 (.76, 1.5).891 (.57, 1.4)
       Gender (male as reference)1.32
      P ≤ .05.
      (1.0, 1.8)
      1.15 (.84, 1.6)1.31 (.98, 1.8)
       Education (>16 as reference)1.36
      P ≤ .05.
      (1.0, 1.8)
      1.46
      P ≤ .05.
      (1.1, 2.0)
      1.16 (.88, 1.5)1.28 (.93, 1.8)
       Social class (high as reference)1.45
      P ≤ .05.
      (1.1, 1.9)
      1.54
      P ≤ .05.
      (1.1, 2.1)
      1.47
      P ≤ .05.
      (1.1, 1.9)
      1.47
      P ≤ .05.
      (1.1, 2.1)
      1.14 (.85, 1.5)1.19 (.86, 1.7)
      Pain
       Pain Intensity1.23
      P ≤ .001.
      (1.2, 1.3)
      1.24
      P ≤ .001.
      (1.2, 1.3)
      1.15
      P ≤ .001.
      (1.1, 1.2)
      1.11
      P ≤ .05.
      (1.0, 1.2)
      1.12
      P ≤ .05.
      (1.0, 1.2)
      1.09 (1.0, 1.2)
       RMDQ1.08
      P ≤ .001.
      (1.1, 1.1)
      1.09
      P ≤ .001.
      (1.1, 1.1)
      1.04
      P ≤ .05.
      (1.0, 1.1)
      1.05
      P ≤ .05.
      (1.0, 1.1)
      1.02 (.98, 1.1)1.01 (.97, 1.1)
       Pain duration (<3 years as reference)1.51
      P ≤ .05.
      (1.1, 2.0)
      1.82
      P ≤ .001.
      (1.3, 2.5)
      1.15 (.85, 1.5)1.34 (.96, 1.9)
       Leg pain (no as reference)1.79
      P ≤ .001.
      (1.3, 2.4)
      2.10
      P ≤ .001.
      (1.5, 3.0)
      1.09 (.79, 1.5)1.25 (.85, 1.9)
       Distal pain (no as reference)1.27
      P ≤ .05.
      (1.0, 1.5)
      1.40
      P ≤ .05.
      (1.2, 1.6)
      1.07 (.72, 1.3)1.20 (.85, 1.4)
      Psychological
       Anxiety (HADS)1.06
      P ≤ .001.
      (1.0, 1.1)
      1.07
      P ≤ .001.
      (1.0, 1.1)
      .995 (.95, 1.0)1.03 (.98, 1.1)
       Depression (HADS)1.08
      P ≤ .001.
      (1.1, 1.1)
      1.09
      P ≤ .001.
      (1.1, 1.1)
      .989 (.94, 1.0).984 (.92, 1.1)
       TSK, Fear of movement1.05
      P ≤ .001.
      (1.0, 1.1)
      1.05
      P ≤ .001.
      (1.0, 1.1)
      1.00 (.98, 1.0).992 (.96, 1.0)
       CSQ, Catastrophizing1.05
      P ≤ .001.
      (1.0, 1.1)
      1.05
      P ≤ .001.
      (1.0, 1.1)
      1.00 (.98, 1.0)1.01 (.98, 1.1)
       CSQ, Diversion1.02
      P ≤ .05.
      (1.0, 1.0)
      1.02
      P ≤ .05.
      (1.0, 1.0)
      1.00 (.98, 1.0)1.02 (.99, 1.1)
       CSQ, Reinterpretation.995 (.98, 1.0).994 (.97, 1.0)
       CSQ, Cognitive coping.974
      P ≤ .05.
      (.95, .99)
      .973
      P ≤ .05.
      (.95, .99)
      1.02 (.99, 1.1)1.01 (.97, 1.1)
       IPQR, Symptoms1.15
      P ≤ .001.
      (1.1, 1.2)
      1.16
      P ≤ .001.
      (1.1, 1.2)
      1.02 (.95, 1.1).981 (.90, 1.1)
       IPQR, Consequences1.09
      P ≤ .001.
      (1.1, 1.1)
      1.09
      P ≤ .001.
      (1.1, 1.1)
      1.01 (.97, 1.1)1.03 (.97, 1.1)
       IPQR, Cyclical time1.00 (.96, 1.0)1.01 (.97, 1.1)
       IPQR, Emotion representation1.07
      P ≤ .001.
      (1.1, 1.1)
      1.06
      P ≤ .001.
      (1.0, 1.1)
      1.01 (.97, 1.1).952
      P ≤ .05.
      (.91, .99)
      .975 (.94, 1.0)
       IPQR, Illness coherence1.02 (.99, 1.0)1.03
      P ≤ .05.
      (1.0, 1.1)
      1.01 (.98, 1.1)
       IPQR, Personal control.932
      P ≤ .001.
      (.90, .97)
      .914
      P ≤ .001.
      (.88, .95)
      .974 (.93, 1.0).983 (.93, 1.0)
       IPQR, Treatment control.930
      P ≤ .001.
      (.90, .97)
      .900
      P ≤ .001.
      (.86, .94)
      1.03 (.97, 1.1)1.01 (.95, 1.1)
       IPQR, Timeline acute-chronic1.07
      P ≤ .001.
      (1.0, 1.1)
      1.09
      P ≤ .001.
      (1.1, 1.1)
      1.04
      P ≤ .05.
      (1.0, 1.1)
      1.06
      P ≤ .05.
      (1.0, 1.1)
      1.04
      P ≤ .05.
      (1.0, 1.1)
      1.06
      P ≤ .001.
      (1.0, 1.1)
       IPQR, Psychological attributions1.02 (.99, 1.1)1.04 (1.0, 1.1)1.02 (.96, 1.1)
       IPQR, Risk factors1.01 (.97, 1.0)1.02 (.99, 1.1)
       IPQR, Immunity1.04 (.98, 1.1)1.08
      P ≤ .05.
      (1.0, 1.2)
      1.00 (.90, 1.1)
       IPQR, Accident/chance1.06 (.99, 1.1)1.10
      P ≤ .05.
      (1.0, 1.2)
      .999 (.92, 1.1)
       Pain self-efficacy.970
      P ≤ .001.
      (.96, .98)
      .967
      P ≤ .001.
      (.96, .98)
      .973
      P ≤ .05.
      (.96, .99)
      .973
      P ≤ .05.
      (.96, .99)
      .990 (.98, 1.0).989 (.97, 1.0)
       Active coping1.05 (.91, 1.2).996 (.85, 1.2)
       Passive coping1.23
      P ≤ .001.
      (1.1, 1.4)
      1.31
      P ≤ .001.
      (1.2, 1.5)
      1.06 (.95, 1.2)1.16
      P ≤ .05.
      (1.0, 1.3)
      1.11 (.98, 1.3)
      Occupational
       Employed (reference)
      Employed, reduced duties (LBP)1.86
      P ≤ .05.
      (1.3, 2.6)
      1.87
      P ≤ .05.
      (1.2, 2.8)
      1.13 (.75, 1.7)1.26 (.78, 2.0)
      Unemployed1.89
      P ≤ .001.
      (1.3, 2.7)
      2.18
      P ≤ .001.
      (1.5, 3.2)
      1.15 (.75, 1.8)1.37 (.84, 2.2)
      Unemployed (LBP)2.93
      P ≤ .001.
      (2.0, 4.3)
      3.86
      P ≤ .001.
      (2.6, 5.8)
      1.02 (.61, 1.7)1.24 (.70, 2.2)
      P ≤ .05.
      ∗∗ P ≤ .001.

      Demographic Domain

      Univariate tests show that age did not predict outcome at either 6 months or 5 years; female gender predicted poor outcome at 6 months but not at 5 years. Reporting having no further education beyond the age of 16 years and lower social class were each associated with an increased risk of poor outcome at 6 months and at 5 years. In the demographic multivariable model, only lower social class remained predictive (increased risk of clinically significant LBP at 6 months and at 5 years).

      Physical Domain

      All variables in the physical domain significantly predicted risk of poor outcome in the univariate tests, but only pain intensity and disability remained after mutual adjustment for the intercorrelation between all the other variables within the pain domain model.

      Psychological Domain

      Most of the variables within the psychological domain predicted clinically significant LBP status at 6 months and at 5 years. However, following adjustment in the psychological multivariable model, only a stronger baseline perception by the patient that his or her pain will last a long time (IPQR timeline acute-chronic variable) and lower pain self-efficacy (confidence in his or her ability to get on with life despite the pain) independently predicted clinically significant LBP at 6 months and at 5 years. A further 2 variables were predictive of the 5-year outcome but not the 6-month outcome: lower levels of emotional representations of one's back pain problem (IPQR–emotional representations) and higher passive coping.

      Occupational Domain

      The occupational variable consisted of a single question, so no statistical adjustment was required. The model showed that compared to those in employment, persons on reduced duties because of their back problem at baseline, those who were unemployed, and those who were unemployed because of LBP were all at an increased risk of clinically significant LBP at 6 months and at 5 years.

      Final Multivariable Model

      In the final multivariable model (Table 2), which combined significant factors remaining from each domain, baseline pain intensity significantly predicted 6-month outcome (RR = 1.12, 95% CI = 1.03–1.20) and showed a trend for 5-year outcome (RR = 1.09, 95% CI = .99–1.20), with a 12% and 9% increase in risk, respectively, per unit change on the pain intensity score. To give clinical understanding to this per unit change in score, a person with a one-third higher score of baseline pain intensity would have a 44% increase in risk of poor outcome at 6 months and a 33% increase in risk at 5 years. In addition, the IPQR timeline acute-chronic variable (ie, a measure of how long the person believes his or her back pain will last) showed a statistically significant increase in risk of 4% at 6 months (RR = 1.04, 95% CI = 1.01–1.07), and 6% at 5 years per unit change on the scale score (RR = 1.06, 95% CI = 1.03–1.09). Again to show clinical interpretation to this result, a person with a one-third higher score on the timeline scale at baseline would be at an increase of 32% in risk of poor outcome at 6 months and a 48% increase in risk at 5 years.

      Responder Nonresponder Analysis

      Responders at 6 months were older (mean, 45.5 vs 42.2 years) and more likely to be female (61 vs 56%) than nonresponders (see Foster et al
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      for full details of baseline and 6-month stages). Similarly, responders at 5 years were older (mean 47.4 vs 43.9 years), but there was no difference in gender. There were no reported differences between responders and nonresponders on the main outcomes (see Appendix 1 for full tests).

      Discussion

      In a representative sample of UK primary care patients with LBP followed up for 5 years, elevated pain intensity around the time of the index healthcare visit and patients' perception that their back problems will last a long time were significant predictors of poor outcome in both the short and long term.

      Comparison With Existing Literature

      This study confirms the findings from other prognostic studies that baseline pain intensity is a key predictor of future pain and disability status,
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      and it shows similar findings to other studies within primary care settings in respect to patient beliefs about the longevity of their LBP.
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      • Henschke N.
      • Maher C.G.
      • Refshauge K.M.
      • Herbert R.D.
      • Cumming R.G.
      • Bleasel J.
      • York J.
      • Das A.
      • McAuley J.H.
      Prognosis in patients with recent onset low back pain in Australian primary care: Inception cohort study.
      This study is the first to demonstrate this over a long time frame, and importantly shows the stability of these key prognostic factors over both the short- and long term. Two studies of primary care back pain patients, 1 of which combined baseline data from this cohort,
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      showed that employment factors (eg, being unemployed, absence from work) predicted poor outcome at 12 months.
      • Dunn K.M.
      • Jordan K.P.
      • Croft P.R.
      Contributions of prognostic factors for poor outcome in primary care low back pain patients.
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      However, these studies utilized outcomes on the basis of disability,
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
      or a more severe level of pain and disability compared to this study,
      • Dunn K.M.
      • Jordan K.P.
      • Croft P.R.
      Contributions of prognostic factors for poor outcome in primary care low back pain patients.
      and it may be that employment factors play a greater role in predicting more severe outcome. One long-term study (Burton et al
      • Burton A.K.
      • McClune T.D.
      • Clarke R.D.
      • Main C.J.
      Long-term follow-up of patients with low back pain attending for manipulative care: Outcomes and predictors.
      ) compared prognostic factors for LBP disability at 1 and 4 years. They report that fear avoidance, catastrophizing, and passive coping predicted poor outcome at 1 year but that depression and pain intensity predicted poor outcome at 4 years, and suggest that poor initial coping leads to longer-term depression. The current study does not reflect these findings, as there was little difference in factors predicting short-term (6 months) and long-term (5 years) outcome. The differences in findings between the current study and Burton et al might be explained by case mix and design. For example, the Burton et al study recruited patients seeking osteopathic treatment in private practice settings, with patients receiving at least 6 weeks of manipulative therapy after consultation. It may be that those who were initially depressed were less likely to respond to treatment, and the difference in the predictive factors for long-term disability may be an artifact of this response to treatment, rather than actual differences between individuals at baseline. One other study, albeit with only 12-month follow-up, that has design similarities to our study is done by Henschke et al.
      • Henschke N.
      • Maher C.G.
      • Refshauge K.M.
      • Herbert R.D.
      • Cumming R.G.
      • Bleasel J.
      • York J.
      • Das A.
      • McAuley J.H.
      Prognosis in patients with recent onset low back pain in Australian primary care: Inception cohort study.
      They studied primary care patients with LBP, included similar domain analysis, and used a measure similar to this study's timeline variable: a patient self-rated measure of how long his or her pain or disability will persist. They found that a patient's belief in the persistence of his or her back pain predicted poor outcome, alongside pain intensity, depression, and compensation claim status. Taken together with the current study's findings, this strengthens the conclusion that patients' beliefs about their back pain are important and robust prognostic markers in both the short and long term. A recent review on psychosocial risk factors for chronic LBP by Ramond et al
      • Ramond A.
      • Bouton C.
      • Richard I.
      • Roquelaure Y.
      • Baufreton C.
      • Legrand E.
      • Huez J.F.
      Psychosocial risk factors for chronic low back pain in primary care—A systematic review.
      reported that expectations about recovery conferred a more consistent risk factor than other psychological factors in predicting outcome; however, the included studies showing this effect had follow-up durations of up to 12 months. Our study has shown that such expectations are exerting a relatively strong influence (compared with a wide range of psychological and demographic factors) on long-term outcome 5 years later.

      Strengths and Weaknesses

      Major strengths of this study are the large sample of primary care patients with LBP, long-term follow-up of 5 years, the comparison between predictive factors over the short and long term, and the wide range of predictors measured using validated instruments.
      • Hestbaek L.
      • Leboeuf-Yde C.
      • Manniche C.
      Low back pain: What is the long term course? A review of studies of general patient populations.
      • Mallen C.D.
      • Peat G.
      • Thomas E.
      • Dunn K.M.
      • Croft P.R.
      Prognostic factors for musculoskeletal pain in primary care: A systematic review.
      In addition, this study concurs with a comprehensive review of prognostic studies on the high level of chronicity over time
      • Hestbaek L.
      • Leboeuf-Yde C.
      • Manniche C.
      Low back pain: What is the long term course? A review of studies of general patient populations.
      ; however, we do acknowledge that a proportion of the baseline population would have changed in pain and disability status in the approximate 2 weeks between index consultation and response to baseline questionnaire, and this may have influenced the effect estimates.
      Our results are bounded by the analysis model we have chosen. We chose a model that helped manage the large amount of predictor variables entered in the analysis by separating them into clinically meaningful domains of influence (ie, demographics, physical, psychological, occupational). This then allowed for us to select, statistically, the key predictor variables from each domain before moving them onto a final multivariable model. However, it is possible that predictor variables within one domain could have influenced the coefficient effects of another predictor variable in another domain had they been allowed to model together, for example, in a stepwise regression procedure. However, there are limitations with stepwise modeling; multiple testing may be increased (especially if the number of steps is large) and predictors would be selected solely on the condition of the previous step, which can lead to idiosyncratic models that are hard to replicate.
      • Thompson B.
      Stepwise regression and stepwise discriminant analysis need not apply here: A guidelines editorial.
      Although we included a wide range of predictive factors, other factors not included here, such as length of unemployment and compensation status for LBP injury, have been shown to influence prognosis of LBP.
      • Henschke N.
      • Maher C.G.
      • Refshauge K.M.
      • Herbert R.D.
      • Cumming R.G.
      • Bleasel J.
      • York J.
      • Das A.
      • McAuley J.H.
      Prognosis in patients with recent onset low back pain in Australian primary care: Inception cohort study.
      • Steenstra I.A.
      • Verbeek J.H.
      • Heymans M.W.
      • Bongers P.M.
      Prognostic factors for duration of sick leave in patients sick listed with acute low back pain: A systematic review of the literature.
      It may also be the case that predictors of LBP outcome are not fixed at specific time points (ie, at baseline in this study) but are more fluid in nature and change and evolve during different periods of the person's experience of LBP. Future studies with more frequent follow-ups will be better placed to identify such patterns. There was no evidence of bias caused by participant dropout on the prognostic factors or the outcome; therefore, our findings are unlikely to be substantially influenced by loss to follow-up.
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      • Grotle M.
      • Foster N.E.
      • Dunn K.M.
      • Croft P.
      Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.

      Clinical and Research Relevance

      This study has shown that pain intensity predicts future pain and disability even after 5 years, and it confirms that pain relief is an important target not only in the initial management of the symptom but for the potential contribution to long-term improvement. Moreover, our results show that a strong personal belief that back pain will last a long time predicts clinically significant LBP independently of a wide range of other prognostic factors, including pain intensity, and additionally this belief appears to have enduring strength predicting both short- and long-term poor outcome. Collectively, the assessment of patients' pain intensity and this timeline belief could constitute important prognostic markers for the clinician, which could now be tested for their usefulness and impact in clinical practice. However, as outlined in the introduction of this paper, there is large variation within prognostic studies as to what prognostic factors are the most important. Consequently, this has led to debate and conflict between studies. Current thought suggests a more complex interaction between factors rather than singular predictive factors that influence the patient through time.
      • Nicholas M.K.
      • Linton S.J.
      • Watson P.J.
      • Main C.J.
      Early identification and management of psychological risk factors (“yellow flags”) in patients with low back pain: A reappraisal.
      In agreement, Ramond et al
      • Ramond A.
      • Bouton C.
      • Richard I.
      • Roquelaure Y.
      • Baufreton C.
      • Legrand E.
      • Huez J.F.
      Psychosocial risk factors for chronic low back pain in primary care—A systematic review.
      recently described psychosocial risk factors associated with LBP as momentary indicators that form part of a dynamic of the person in the context of his or her whole experience prior to, during, and after the pain episode. There is clearly a need now to investigate how prognostic factors work together, hypothesized and tested using more sophisticated techniques (eg, structured equation modeling), within longitudinal designs.
      Although the identification of a prognostic marker is important to characterize those at higher risk, consideration should also be given to the potential for modification of such prognostic markers in the effort to alter patients' long-term outcome.
      • Hayden J.A.
      • Dunn K.M.
      • van der Windt D.A.
      • Shaw W.S.
      What is the prognosis of back pain?.
      Previously within this cohort, it was shown that change in the patient's global rating of his or her LBP was associated with change in the patient's timeline belief (ie, as pain reduced, so did the strength of belief of the patient that his or her back pain would last a long time), suggesting that a reduction in pain may lead to a reduction in the strength of the belief.
      • Foster N.E.
      • Bishop A.
      • Thomas E.
      • Main C.
      • Horne R.
      • Weinman J.
      • Hay E.
      Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
      Nevertheless, comparable belief constructs, such as a lower expectations about recovery (ie, the patient believes he or she will not get better), are associated with lower adherence to advice and treatment for LBP and may hinder recovery and maintain pain and disability.
      • Main C.J.
      • Foster N.
      • Buchbinder R.
      How important are back pain beliefs and expectations for satisfactory recovery from back pain.
      It may be that there is a reciprocal relationship between pain and the belief held by the patient, but clearly more research is required to investigate these interactions. However, there is evidence that such beliefs are modifiable.
      • Hale E.D.
      • Treharne G.J.
      • Kitas G.D.
      The common sense model of self-regulation of health and illness: How can we use it to understand and respond to our patients' needs?.
      At present, the path of usual care for LBP (eg, NICE guidelines in the UK

      Savigny P, Kuntze S, Watson P, Underwood M, Ritchie G, Cotterell M, Hill D, Browne N, Buchanan E, Coffrey P, Dixon P, Drummond C, Flanagan M, Greenough C, Griffiths M, Halliday-Bell J, Hettinga D, Vogel S, Walsh D. Low back pain: Early management of persistent non-specific low back pain. National Collaborating Centre for Primary Care and Royal College of General Practitioners, London (14), 2009

      ) initially focuses on advice followed by physically based treatments, and not until later are psychological treatments advocated. Our findings add to those from other studies and indicate that a combined approach (ie, pain management and addressing patient's beliefs) early in the treatment process may be beneficial in averting a maladaptive and potentially harmful belief for patients with LBP.

      Conclusion

      From a wide range of prognostic factors, the patient's baseline pain intensity and a belief that his or her LBP will last a long time increased the risk of clinically significant LBP in both the short and long term. Better pain management coupled with specific identification and modification of patients' perceptions of the future of their back problem are clear targets for clinical interventions.

      Acknowledgments

      The authors thank Professor Peter Croft, Professor of Primary Care Epidemiology at the Arthritis Research UK Primary Care Centre, Keele University, for his critical comments during the drafting of this manuscript. They also thank the administrative and health informatics staff at the Arthritis Research UK Primary Care Centre and the Keele General Practice Partnership. Thanks are also extended to the doctors, staff, and patients of the 8 participating general practices.

      Appendix 1

      Prognostic Indicators of Low Back Pain in Primary Care: 5-Year Prospective Study

      Distributional checks and comparisons between those traced and not traced and between responders and nonresponders for the follow-up periods of 6 months and 5 years.
      Distributional checks were carried out on continuous indicator variables to determine which showed non-normal distributions, indicating nonparametric testing for response and nonresponse and traced/nontraced groupings; chi-square testing was carried out for all categorical indicator variables. For differences between those who were traced (n = 698) and those not traced (n = 112) at the 5-year follow-up, analysis revealed a significant difference in age between those who were traced and those not traced. Those who were not traced were significantly younger in age (41.4 vs 46.3). There were no other significant differences found between traced and nontraced patients.
      Comparing responders (n = 488) with nonresponders (n = 210) showed significant differences on age; nonresponders were significantly younger (43.3 vs 47.4). Nonresponders scored lower on the CSQ diversion score, higher on the IPQR illness coherence score and higher on the IPQR accident chance score. There were no other differences between responders and nonresponders.
      Appendix Table 1aTest of Baseline Characteristics for Those Traced and Those Not Traced at Follow-Up
      VariableTraced (n = 698)Not Traced (n = 112)Significance
      MedianIQRMeanSDMedianIQRMeanSD
      Age471446.279.4241.51941.4211.12<.001
      Significant difference.
      RMDQ898.725.79888.905.52.473
      Pain intensity3.6733.942.274.333.084.272.41.301
      Depression666.324.12676.384.09.715
      Anxiety868.104.42868.454.43.277
      TSK39839.246.88417.1440.296.32.204
      Catastrophizing8109.347.4791010.407.89.176
      Diversion161115.487.97161415.948.72.608
      Reinterpretation697.396.496118.177.49.551
      Cognitive coping17816.776.1816.50916.646.37.549
      Symptoms434.092.261434.362.07.130
      Consequences17917.335.4617817.325.29.818
      Timeline cyclic14513.103.2813.504.2519.923.37.751
      Emotional16716.575.22177.2516.884.93.483
      Illness coherence12713.405.02128.2513.595.43.477
      Personal control21621.033.6821.5621.054.22.542
      Treatment control17517.203.2317416.843.57.056
      Timeline21919.775.7921920.176.18.550
      Attributions (psychological)12511.723.9812512.044.27.536
      Risk factors15614.944.0115.55.5515.114.08.698
      Immunity635.221.91635.191.82.963
      Accident625.971.97626.251.94.337
      Self efficacy4122.2236.6314.25382238.0513.28.482
      Active coping121.25.92111.25.88.851
      Passive coping222.181.36222.351.31.162
      Categorical variables (Pearson chi-square tests)
       Gender.342
       Pain duration over 3 years.772
       Education level.073
       Social class.178
       Pain medication.969
       Consultation rate.111
      Significant difference.
      Appendix Table 1bTest of Baseline Characteristics for Responders and Nonresponders
      VariableResponders (n = 488)Nonresponders (n = 210)Significance
      MedianIQRMeanSDMedianIQRMeanSD
      Age491447.369.07441742.8910.31<.001
      Significant difference.
      RMDQ898.685.85898.855.60.242
      Pain intensity3.6733.962.3143.334.022.26.558
      Depression666.264.23666.433.94.133
      Anxiety868.034.4386.758.334.41.323
      TSK39839.096.92407.5639.846.63.200
      Catastrophizing89.29.257.47910.009.877.64.145
      Diversion161115.877.99151215.048.18.041
      Significant difference.
      Reinterpretation697.396.52610.757.656.81.891
      Cognitive coping17816.866.0517916.606.43.361
      Symptoms434.112.25444.162.21.972
      Consequences17817.295.4517817.385.41.268
      Timeline cyclic14513.003.2514513.193.35.283
      Emotional16716.555.2717716.725.05.265
      Illness coherence12713.134.9413813.895.25.01
      Significant difference.
      Personal control21621.173.72215.7520.833.81.153
      Treatment control17517.193.1817517.103.44.983
      Timeline21819.995.66211019.576.11.386
      Attributions (psych)12511.693.9212511.884.15.575
      Risk factors15615.023.9915614.884.07.909
      Immunity635.221.88635.191.91.887
      Accident625.871.89636.231.96.04
      Significant difference.
      Self-efficacy412139.1014.02402337.6914.22.184
      Active coping121.23.92111.27.91.358
      Passive coping222.211.34222.201.373.668
      Categorical variables (Pearson chi-square tests)
       Gender.179
       Pain duration over 3 years.499
       Education level.649
       Socioeconomic status.703
       Pain medication.964
       Consultation rate.248
      Significant difference.

      References

        • Artus M.
        • van der Windt D.A.
        • Jordan K.P.
        • Hay E.M.
        Low back pain symptoms show a similar pattern of improvement following a wide range of primary care treatments: A systematic review of randomized clinical trials.
        Rheumatology (Oxford). 2010; 49: 2346-2356
        • Bowling A.
        Research Methods in Health.
        Open University Press, Buckingham1997
        • Brown G.K.
        • Nicassio P.M.
        Development of a questionnaire for the assessment of active and passive coping strategies in chronic pain patients.
        Pain. 1987; 31: 53-64
        • Burton A.K.
        • McClune T.D.
        • Clarke R.D.
        • Main C.J.
        Long-term follow-up of patients with low back pain attending for manipulative care: Outcomes and predictors.
        Man Ther. 2004; 9: 30-35
        • Dunn K.M.
        • Croft P.R.
        The importance of symptom duration in determining prognosis.
        Pain. 2006; 121: 126-132
        • Dunn K.M.
        • Jordan K.
        • Croft P.R.
        Characterizing the course of low back pain: A latent class analysis.
        Am J Epidemiol. 2006; 163: 754-761
        • Dunn K.M.
        • Jordan K.P.
        • Croft P.R.
        Recall of medication use, self-care activities and pain intensity: A comparison of daily diaries and self-report questionnaires among low back pain patients.
        Primary Health Care Res Develop. 2010; 11: 93-102
        • Dunn K.M.
        • Jordan K.P.
        • Croft P.R.
        Contributions of prognostic factors for poor outcome in primary care low back pain patients.
        Eur J Pain. 2011; 15: 313-319
        • Elliott A.M.
        • Smith B.H.
        • Smith W.C.
        • Chambers W.A.
        Changes in chronic pain severity over time: The Chronic Pain Grade as a valid measure.
        Pain. 2008; 88: 303-308
        • Foster N.E.
        • Bishop A.
        • Thomas E.
        • Main C.
        • Horne R.
        • Weinman J.
        • Hay E.
        Illness perceptions of low back pain patients in primary care: What are they, do they change and are they associated with outcome?.
        Pain. 2008; 136: 177-187
        • George C.
        The six-month incidence of clinically significant low back pain in the Saskatchewan adult population.
        Spine. 2002; 27: 1778-1782
        • Grotle M.
        • Foster N.E.
        • Dunn K.M.
        • Croft P.
        Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care?.
        Pain. 2010; 151: 790-797
        • Hale E.D.
        • Treharne G.J.
        • Kitas G.D.
        The common sense model of self-regulation of health and illness: How can we use it to understand and respond to our patients' needs?.
        Rheumatology (Oxford). 2007; 46: 904-906
        • Harland N.J.
        • Georgieff K.
        Development of the Coping Strategies Questionnaire 24, a clinically utilitarian version of the Coping Strategies Questionnaire.
        Rehabil Psychol. 2003; 48: 296-300
        • Hayden J.A.
        • Dunn K.M.
        • van der Windt D.A.
        • Shaw W.S.
        What is the prognosis of back pain?.
        Best Pract Res Clin Rheumatol. 2010; 24: 167-179
        • Henschke N.
        • Maher C.G.
        • Refshauge K.M.
        • Herbert R.D.
        • Cumming R.G.
        • Bleasel J.
        • York J.
        • Das A.
        • McAuley J.H.
        Prognosis in patients with recent onset low back pain in Australian primary care: Inception cohort study.
        BMJ. 2008; 337: a171
        • Hestbaek L.
        • Leboeuf-Yde C.
        • Manniche C.
        Low back pain: What is the long term course? A review of studies of general patient populations.
        Eur Spine J. 2003; 12: 149-165
        • Hoy D.
        • Brooks P.
        • Blyth F.
        • Buchbinder R.
        The epidemiology of low back pain.
        Best Pract Res Clin Rheumatol. 2010; 24: 769-781
        • Katz M.H.
        Multivariable Analysis: A Practical Guide for Clinicians.
        2nd ed. Cambridge University Press, Cambridge2006
        • Kori S.H.
        • Miller R.P.
        • Todd D.D.
        Kinesiophobia: A new view of chronic pain behaviour.
        Pain Manag. 1990; 3: 35-43
        • Leboeuf-Yde C.
        • Nielsen J.
        • Kyvik K.O.
        • Fejer R.
        • Hartvigsen J.
        Pain in the lumbar, thoracic or cervical regions: Do age and gender matter? A population-based study of 34,902 Danish twins 20-71 years of age.
        BMC Musculoskelet Disord. 2009; 10: 39
      1. Lumley T, Kronmal R, Shuangge M: Relative risk regression in medical research: Models, contrasts, estimators and algorithms. UW Biostatistics Working Paper Series, paper 293. Available at: http://www.bepress.com/uwbiostat/paper293. Accessed February 2013

        • Main C.J.
        • Foster N.
        • Buchbinder R.
        How important are back pain beliefs and expectations for satisfactory recovery from back pain.
        Best Pract Res Clin Rheumatol. 2010; 24: 205-217
        • Mallen C.D.
        • Peat G.
        • Thomas E.
        • Dunn K.M.
        • Croft P.R.
        Prognostic factors for musculoskeletal pain in primary care: A systematic review.
        Br J Gen Pract. 2007; 57: 655-661
        • Manchikanti L.
        • Singh V.
        • Datta S.
        • Cohen S.P.
        • Hirsch J.A.
        Comprehensive review of epidemiology, scope, and impact of spinal pain.
        Pain Physician. 2009; 12: E35-E70
        • Moss-Morris R.W.
        • Weinman J.
        • Petrie K.J.
        • Horne R.
        • Cameron L.D.
        • Buick D.
        The revised Illness Perception Questionnaire (IPQ-R).
        Psychol Health. 2002; 17: 1-16
        • Nicholas M.K.
        The pain self-efficacy questionnaire: Taking pain into account.
        Eur J Pain. 2007; 11: 153-163
        • Nicholas M.K.
        • Linton S.J.
        • Watson P.J.
        • Main C.J.
        Early identification and management of psychological risk factors (“yellow flags”) in patients with low back pain: A reappraisal.
        Phys Ther. 2011; 91: 737-753
        • Office for National Statistics
        Standard occupational classification. The Codingindex, vol. 2, The Stationary Office, Version 1.1, The National Statistics Socio-economic Classification User Manual.
        Office for National Statistics, London2002
        • Papageorgiou A.C.
        • Croft P.R.
        • Ferry S.
        • Jayson M.I.
        • Silman A.J.
        Estimating the prevalence of low back pain in the general population. Evidence from the South Manchester Back Pain Survey.
        Spine. 1995; 20: 1889-1894
        • Picavet H.S.J.
        • Schouten J.S.
        Musculoskeletal pain in the Netherlands: Prevalences, consequences and risk groups, the DMC3-study.
        Pain. 2003; 102: 167-178
        • Pincus T.
        • Burton A.K.
        • Vogel S.
        • Field A.P.
        A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain.
        Spine. 2002; 27: E109-E120
        • Porcheret M.
        • Hughes R.
        • Evans D.
        • Jordan K.
        • Whitehurst T.
        • Ogden H.
        • Croft P.
        North Staffordshire general practice research network. Data quality of general practice electronic health records: The impact of a program of assessments, feedback, and training.
        J Am Med Inform Assoc. 2004; 11: 78-86
        • Ramond A.
        • Bouton C.
        • Richard I.
        • Roquelaure Y.
        • Baufreton C.
        • Legrand E.
        • Huez J.F.
        Psychosocial risk factors for chronic low back pain in primary care—A systematic review.
        Fam Pract. 2011; 28: 12-21
        • Roland M.
        • Morris R.
        A study of the natural history of back pain. Part I: Development of a reliable and sensitive measure of disability in low-back pain.
        Spine. 1983; 8: 141-144
      2. Savigny P, Kuntze S, Watson P, Underwood M, Ritchie G, Cotterell M, Hill D, Browne N, Buchanan E, Coffrey P, Dixon P, Drummond C, Flanagan M, Greenough C, Griffiths M, Halliday-Bell J, Hettinga D, Vogel S, Walsh D. Low back pain: Early management of persistent non-specific low back pain. National Collaborating Centre for Primary Care and Royal College of General Practitioners, London (14), 2009

        • Smith B.H.
        • Penny K.I.
        • Purves A.M.
        • Munro C.
        • Wilson B.
        • Grimshaw J.
        • Chambers A.W.
        • Cairns Smith W.
        The Chronic Pain Grade questionnaire: Validation and reliability in postal research.
        Pain. 1997; 71: 141-147
        • Steenstra I.A.
        • Verbeek J.H.
        • Heymans M.W.
        • Bongers P.M.
        Prognostic factors for duration of sick leave in patients sick listed with acute low back pain: A systematic review of the literature.
        Occup Environ Med. 2005; 62: 851-860
        • Thompson B.
        Stepwise regression and stepwise discriminant analysis need not apply here: A guidelines editorial.
        Educ Psychol Meas. 1995; 55: 525-534
        • Thompson M.L.
        • Myers J.E.
        • Kriebel D.
        Prevalence odds ratio or prevalence ratio in the analysis of cross sectional data: What is to be done?.
        Occup Environ Med. 1998; 55: 272-277
        • Von K.M.
        • Miglioretti D.L.
        A prognostic approach to defining chronic pain.
        Pain. 2005; 117: 304-313
        • Von K.M.
        • Ormel J.
        • Keefe F.J.
        • Dworkin S.F.
        Grading the severity of chronic pain.
        Pain. 1992; 50: 133-149
        • Zadurian N.
        The role of coping styles and strategies in patients consulting for low back pain in primary care.
        (Thesis (PhD)) University of Keele, 2012
        • Zigmond A.S.
        • Snaith R.P.
        The Hospital Anxiety and Depression Scale.
        Acta Psychiatr Scand. 1983; 67: 361-370