Forecasting is tough work. Whether one is a meteorologist or a stock broker, it takes guts to make a prediction. The same applies to making predictions in the physical therapy profession.

The beauty of a prediction is that it is testable. We can look at a set of data, make a reasonable guess about what will happen, and see what actually happens. All too often, health professionals avoid gathering enough data to even make a prediction. If you’re a therapist, then perhaps you’re tempted to get started on treatment because you’re afraid of irritating your patient with a lengthy evaluation. Perhaps you doubt whether standardized outcome measures really make a difference. Perhaps you’re just so overwhelmed by the sea of available measures that you won’t even test the water.

But if you treat low back pain, then one short, simple measure demands your attention. It’s called the STarT Back Screening Tool, and it’s changing the physical therapy profession.

The STarT Back Tool (abbreviated from Subgroups for Targeted Treatment Back Screening Tool) consists of nine items. Here is a printable version. It includes items about several aspects of low back pain: referred leg pain, comorbid pain, disability (2 items), bothersomeness, catastrophizing, fear, anxiety, and depression. The latter five items comprise a psychosocial subscale. Scores range from 0-9, and a score falls within one of three subgroups based on “risk” for poor prognosis. Low risk scores range from 0 to 3. If someone scores 4 or 5 on the psychosocial subscale, then they are classified as high risk. All other scores are classified as medium risk. 

The goal of the STarT Back Tool was to aid decision-making in primary care. Those considered high risk for poor prognosis may be identified and referred to other services — particularly psychological services — while those considered low risk wouldn’t receive the same treatment. This type of stratified care lends itself well to empirical research, because patients could be randomized into either standard care or stratified care after classification into their respective risk-based subgroups.

The group at Keele University that designed the Tool conducted exactly this experiment. Hill et al. (2011) found that those receiving stratified care showed a greater reduction of disability compared to standard care at 4 months and 12 months (with Cohen’s d of 0.32 and 0.19, respectively, and p < .01 for both time points). What is more, those receiving stratified care achieved a greater mean increase in generic health benefit and greater cost savings compared to the standard care group at 12 months.

There were several differences between the stratified care model and the standard care model. The main distinguishing feature of the stratified care group was high risk patients received psychologically-informed physical therapy, for which the physiotherapists received nine days of training. Also, those in the low risk group received only one visit in the stratified care group, consisting primarily of education and encouragement to exercise and to return to work.

Using the STarT Back Tool in primary care is one thing, but what about in an outpatient physical therapy setting?

At the University of Florida, Jason Beneciuk and Steve George (2015) conducted a study similar to that of Hill et al. (2011), but with a twist. Whereas the Keele University group provided extensive training of therapists, Beneciuk & George provided only 8 hours of training and did not randomly assign treatment based on score on the STarT Back Tool. Instead, Beneciuk & George randomized physical therapists into either a group receiving no training or a group receiving the 8-hour training about psychologically-informed physical therapy and pain neuroscience. The training discussed activity-based interventions such as graded activity/exposure and educational-based interventions that incorporate explanation of basic pain neuroscience.

Beneciuk & George found that those receiving stratified care showed greater improvements compared to standard care at 4 weeks for both pain and disability (with Cohen’s d of 0.4 and 0.76, respectively, and p <= 0.01 for both). What really makes these between-group differences fascinating is that the researchers didn’t actually control the type of interventions. Instead, they merely provided education to physical therapists and let the PTs be PTs.

Based on the data, though, we can’t rule out the possibility that educating PTs about pain neuroscience and psychology could lead to greater therapeutic alliance. Whether or not this study is a victory for psychologically-informed physical therapy, it certainly appears to be a victory both for increased training for PTs and for increased screening using the STarT Back Tool.

What is more, The STarT Back Tool is more predictive of chronic low back pain than are physiologic measures, demonstrates acceptable reliability and validity in several languages, and is mentioned in the clinical practice guidelines for low back pain by the Orthopaedic Section of the APTA. A recent preliminary evaluation even shows promise for modifying the STarT Back Tool to assess other pain conditions.

In an outpatient orthopaedic setting, I’m much more comfortable with predicting prognosis after learning about the STarT Back Tool. Still, it takes hard work to know when to modify your treatment and when to refer your patient.  

But maybe forecasting is a little less tough after all.