Table 3.
Factors associated with beliefs about back pain: results from cross‐sectional and longitudinal studies where beliefs about back pain were hypothesized to represent the independent variable
| Article | Sample (country, n, description) | Belief | Dependent variable | Statistical analysis | Results |
|---|---|---|---|---|---|
| Cross‐sectional studies | |||||
| Elfering et al. (2015) | Switzerland, 2507, mixed BP | BBQ | Pain intensity (0–6 scale) | Pearson correlation, cross‐lagged panel regression analysis | Pain intensity associated with BBQ scores in those with BP (β = 0.38, p < 0.001) and in overall sample (β = 0.17, p < 0.001) |
| Linton et al. (2000b) | Sweden, 720, mixed BP | mFABQ‐PA | Membership of moderate pain intensity group (pain ≥5 on 0–10 scale, <30 days sick leave, not on disability pension) | Discriminant analysis | mFABQ‐PA scores associated with moderate pain intensity at first stage of discriminant analysis (p < 0.05) but did not contribute to the final combined model |
| Linton (2005) | Sweden, 521, mixed BP | mFABQ‐PA | Significant pain intensity (pain >6 on 0–10 scale) | Univariate logistic and multiple regression | Significant pain associated with mFABQ‐PA within univariate (OR: 1.72, 95% CI: 1.21–2.46, p < 0.01) but not multiple regression analyses |
| Urquhart et al. (2008) | Australia, 506, mixed BP, female | BBQ | Pain intensity (tertiles based on 0–100 pain rating scores); disability (Chronic Pain Grade) | Univariate and multiple ordinal regression | High pain intensity and worse disability associated with BBQ scores in multiple regression analyses (OR: 0.93, 95% CI: 0.88–0.99, p < 0.05; OR: 0.93, 95% CI: 0.89–0.97, p < 0.001) |
| Leeuw et al. (2007) | Netherlands, 152, current BP (subset of Houben et al., 2013) | TSK‐SV | Disability (Quebec BP Disability Scale) | Pearson correlation, stepwise regression analysis | Disability associated with TSK‐SV after accounting for age, gender, chronicity, catastrophising, pain intensity and baseline disability (β = 0.20, p < 0.01) |
| Buer and Linton (2002) | Sweden, 917, mixed BP (overlaps with Linton, 2000b sample) | mFABQ‐PA | Pain intensity (no, mild, moderate pain); limitations in activities of daily living (within people with BP) | Logistic regression adjusted for catastrophising, smoking, gender | mFABQ‐PA associated with limitations in activities of daily living, but not pain intensity (p < 0.05) |
| Houben et al. (2005) | Netherlands, 1126, mixed BP | TSK‐G | SF‐36: Physical Functioning, Social Functioning, Role Restrictions, Pain, General Health subscales; Pain Catastrophising Scale | Pearson correlation, univariate regression adjusted for gender, age, back pain status | All SF‐36 subscales and pain catastrophising associated with TSK‐G scores (β range for different subscales=−0.27 to 0.21, p < 0.001; β = 0.38, p < 0.001, respectively) |
| Beales et al. (2015) | Australia, 486, BP within preceding month, age ≥49 | BBQ | Disability (Oswestry Disability Index); ‘interferes with normal activities’; ‘interferes with physical activities’; ‘usually causes work absence’; ‘usually use medication for BP’; ‘usually seek professional care for BP’ | Univariate logistic regression analysis and adjusted analyses for current pain level, age, SF‐12, income, employment. | Disability, interference in normal and physical activities, work absence and medication use associated with BBQ in univariate analyses. Only disability and interference in normal activities associated in adjusted analyses (B: −0.37, 95% CI: −0.49 to 0.25, p < 0.001; OR: 0.95, 95% CI: 0.92–0.98, p < 0.01, respectively) |
| Mannion et al. (2009) | Switzerland, 670, current BP and employed (subset of Elfering et al., 2015) | BBQ; FABQ‐PA and Work | Perceived reduction in work productivity; n days missed work | Spearman rank correlation, hierarchical multiple regression | All beliefs correlated with perceived reduction in work productivity and n days absent. Only FABQ‐Work associated with both outcomes within multiple regression (β = 0.26, p < 0.001; β = 0.15, p < 0.001 respectively) |
| Mannion et al. (2013) | Switzerland, 1071, current BP (overlaps with Elfering et al., 2015) | BBQ; FABQ‐PA and Work | Sought healthcare for BP within preceding 4 weeks | Pearson correlation, univariate and multiple logistic regression | All beliefs associated with healthcare‐seeking within univariate analysis. Only FABQ‐Work associated within multiple regression (OR: 1.025, 95% CI: 1.005–1.044, p < 0.05) |
| Szpalski et al. (1995) | Belgium, 2660, history of BP | Statement about whether BP would be lifelong problem. | ‘Sought healthcare’; ‘had bed rest’; ‘took medicine’; ‘had X‐ray’; ‘had surgery’ | Multiple logistic regression | All previous illness behaviours except having had surgery were associated with agreement with belief after accounting for sociodemographic factors and daily BP (p < 0.001) |
| Walker et al. (2004) | Australia, 1228, BP within preceding 6 months | Statement about fear that BP would impair future work capacity. | Sought healthcare in preceding 6 months | Univariate logistic regression | Agreement with belief associated with having sought healthcare (OR: 2.2, 95% CI: 1.7–2.9) |
| Longitudinal studies | |||||
| Linton et al. (2000a) | Sweden, 449, no spinal pain within previous 12 months | mFABQ‐PA | At 12 months: occurrence of pain and test of physical functioning | Logistic regression analyses adjusted for BMI, age, gender, smoking | Above‐median scores on mFABQ‐PA predicted future occurrence of pain (OR: 2.04, 95% CI: 1.19–3.48) and physical functioning (OR: 1.70, 95% CI: 1.12–2.59) |
| Picavet et al. (2002) | Netherlands, 1571, either presence (n = 411) or absence (n = 1160) of BP at baseline | TSK‐G; PCS | At 6 months: presence of BP, BP with limitation in daily activities, BP which has lasted ≥3 months, severe BP (≥5 on 1–10 scale), BP with disability (Quebec BP Disability Questionnaire) | Logistic regression analyses adjusted for baseline pain severity and disability in analysis for those with BP at baseline | Within people without BP at baseline, TSK‐G predicted disabling BP (OR: 3.1, 95% CI: 1.1–8.7) but no other outcomes at follow‐up. Within people who had BP at baseline, TSK‐G and PCS predicted BP at follow‐up (unadjusted), and severe pain, chronic pain, and pain with limitation in daily activity (adjusted). TSK‐G, but not PCS, predicted disabling pain (adjusted) |
| Linton (2005) | Sweden, 372, absence of BP at baseline | mFABQ‐PA | At 6 months: Significant pain intensity (pain >6 on 0–10 scale) | Logistic regression adjusted for catastrophising | mFABQ‐PA did not predict occurrence of future significant pain (OR: 1.48, 95% CI: 0.66–3.32) |
| Elfering et al. (2015) | Netherlands, 1833, mixed BP | BBQ | At 12 months: Pain intensity (0–6 scale) | Cross‐lagged structural equation panel modelling | BBQ predicted future pain intensity in the overall sample (β = 0.6, p < 0.05), in those who had BP (β = 0.11, p < 0.05), who did not exercise (β = 0.15, p < 0.05), and who had BP and did not exercise at baseline (β = 0.29, p < 0.05), but not in those who did not have BP, who exercised, and who had BP and exercised at baseline |
| Elfering et al. (2009) | Netherlands, 264, presence of BP at baseline | BBQ; FABQ‐PA and Work | Weekly pain intensity (0–6 scale), pain frequency, recovery, work impairment | Multi‐level model | Pain intensity predicted by FABQ‐PA and Work; pain frequency predicted by FABQ‐Work; recovery and perceived work impairment predicted by BBQ and FABQ‐Work |