The factor with the largest contribution in this paper – high pain intensity – is theoretically modifiable in primary care, e.g. using analgesic medication or spinal manipulation

(Chou et this website al., 2007). Although such treatments rarely provide complete pain relief, as the risk factor is common (47% of this sample), even slight improvements in pain management leading to a small shift in mean pain levels could have an important influence on the LBP population. Targeting pain may seem obvious, but the fact that many patients still experience pain after primary care management (Hestbaek et al., 2003) indicates room for improvement. Targeting such a common factor may also conflict with the expectation that we should be looking for less common factors to identify the minority who are at risk for long-term problems, but our whole population approach (in this case a primary care population) indicates that the most benefit for the population would be reached by targeting a group of people with a common factor such as pain. This finding should be considered alongside suggestions that a dominant focus on pain as a target for “cure” might mean that back pain is being overtreated (Deyo et al., 2009). However, the ‘overtreatment’ referred to is predominantly BI 6727 epidural steroid injections, opioids and lumbar magnetic resonance imaging, none of which are first line management approaches in primary

care populations (Van Tulder et al., 2006 and Airaksinen et al., 2006). Other interventions may be warranted which are less focused on the pain itself, and which may also reduce pain levels, such as activity-based interventions, DNA ligase work rehabilitation or cognitive behavioural approaches. The factor identified with the next highest contribution – not being in employment – is more problematic within this setting. In occupational settings, enabling return to work in back pain sufferers is commonly addressed (Nguyen

and Randolph, 2007), and our findings justify that priority. However, people without current employment would not be addressed in an occupational setting. In current UK primary care, GPs rarely have any influence over return to work (if employed) or return to employment (if unemployed). Our findings justify the UK government initiative addressing health, work and wellbeing (http://www.workingforhealth.gov.uk/). A multifactorial approach, acknowledging social influences on LBP, would likely also be beneficial in other settings where health care and employment are separated. The PAF calculations are important intervention strategies for LBP in primary care as a whole, as they estimate the relative contribution of various factors to outcome. Studies in LBP usually only present measures of association (RRs, ORs), but these vary in overall contribution according to how common the risk factors are.