Flags as a method of risk identification
The term ‘yellow flags’ has become a familiar term used to describe psychosocial risk factors for chronicity. The original flag system (Kendall et al. 1997), and its later developments (Main et al. 2005; Kendall et al. 2009) is one such approach to risk identification. It has been described as a methodological compromise between the inflexibilities of a purely actuarial model and a purely subjective approach based on clinical judgement (Linton et al. 2005). In this context it has three important features. Firstly, it offers a ‘systems perspective’ and assumes that an adequate understanding of the problem requires consideration of both the injured worker and the individual’s social and occupational context. Secondly, it contains both clinical and occupational elements. Thirdly, it makes an important distinction between the individual’s perception of the situation and the objective features.
The flags approach can be viewed as a conceptual framework potentially capable of including both actuarial data and individually-assessed risk factors informing different types of specifically targeted interventions based on modifiable risk. An attempted mapping of the flags onto the individual predictors of chronic pain and disability is illustrated in Table 11.1. Different flags, and combinations of flags, require different types on interventions, and a multiflag approach appears to lend itself flexibly to both individual clinical decision-making and widescale system applications (Main et al. 2005; Kendall et al. 2009). However, its predictive accuracy in administrative applications will likely be lower than one arising from a purely statistical or ‘actuarial’ approach and will likely result in over-identification of individuals at risk. Further reflections on the linking of screening with targeting are offered below.
As far as individual risk factors for long-term work disability are concerned, Sullivan et al. (2005) in their review found evidence for fear, beliefs in severity of health conditions, and catastrophizing (yellow flags). Nicholas et al. (2011) having reviewed the evidence both for the influence of yellow flags on outcomes in people with acute/subacute LBP, and of yellow flag targeting on outcomes, concluded:
Overall, from the evidence gathered here the studies that target interventions on known psychological risk factors for disability do seem to be reporting more consistently positive results relative to those interventions that either ignore these risk factors or provide omnibus interventions to people regardless of psychological risk factors. It seems that the identification of those with these risk factors is an important precursor to psychological interventions.
Predictors of occupational outcomes have also been investigated. Shaw et al. (2009) in addition to pain severity and level of depressive symptoms also identified workplace factors such as job stress, co-worker support, job dissatisfaction, employer attitudes, job autonomy, and availability of modified work as influences on duration of work disability and return-to-work (RTW) outcomes. Their findings are consistent with an earlier more widespread review of predictors of chronic pain and disability (Waddell et al, 2003). It is sometimes difficult, however, to distinguish clinical outcomes such as increase in activity or postural tolerance from occupational variables such as RTW rates or indices of work capability.
In fact, the COST B13 prevention guidelines (Burton et al. 2004a) offered a distinction between the general population and workers. Despite some variation in the level of evidence available, the similarity among the recommendations is striking.
The guidelines for workers obviously contain a number of recommendations specific to work settings, but otherwise, in terms of the usefulness of information or educational approaches, the lack of support for traditional biomechanical and biomedical approaches, and for traditional clinical interventions is similar. In fact, over the last decade, with the exception of a series of reviews in the Cochrane Library (http://www2.cochrane.org/reviews/) synthesizing the findings of earlier conservative management, and studies of outcome of surgery (which are beyond the remit of this chapter), there appears to have been relatively little new research with clinical outcomes as a primary focus.
Waddell et al. (2003) appraised the evidence for different sorts of predictors of chronic pain and disability. In their first set of evidence tables they summarize the findings from published studies of clinical and psychosocial predictors (Table 11.1) where the strength of evidence and strength of predictors are shown along with a tentative flag assignment in the right-hand column.
Aggregation of such a large set of data, with such a wide variety of specific variables is by nature imprecise, as are the 27 variables under which the predictors have been gathered. Accepting these strictures, it can be seen that there is evidence for an influence of both sociodemographic and personal history variables, but the strongest influences appear to be from yellow or blue flags (most of which are potentially modifiable) and black flags (which are more immutable, although may present opportunities for a ‘systems solution’; equivalent to Sullivan et al.’s (2005) Type-I (individually centred) and Type-II (workplace or system-based) solutions.