Of participants in this study 84% met the classification of ‘poor outcome’ at 4-month follow-up.
•
Discrimination of poor outcome was considered ‘poor’ for all instruments (area under the receiver operating characteristic curve <.7).
•
Decision curve analysis estimated no net benefit of screening over a ‘treat-all’ approach.
•
The small differences in instrument performances are unlikely to be meaningful.
Abstract
Prognostic screening in patients with low back pain (LBP) offers a practical approach to guiding clinical decisions. Whether screening is helpful in secondary care is unclear. This prospective cohort study in adults with LBP placed on outpatient clinic waiting lists, compared the performance of the short-form Orebro Musculoskeletal Pain Screening Questionnaire, the Predicting the Inception of Chronic Pain Tool, and the STarT Back Tool. We assessed predictive validity for outcome at 4-month follow-up, by calculating estimates of discrimination, calibration, and overall performance. We applied a decision curve analysis approach to describe the clinical value of screening in this setting via comparison with a ‘treat-all’ strategy. Complete data were available for 89% of enrolled participants (n = 195). Eighty-four percent reported ‘poor outcome’ at follow-up. The area under the receiver operating characteristic curve (95% confidence interval) was .66 (.54–.78) for the Orebro Musculoskeletal Pain Screening Questionnaire, .61 (.49–.73) for the Predicting the Inception of Chronic Pain Tool, and .69 (.51–.80) for the STarT Back Tool. All instruments were miscalibrated and underestimated risk. The decision curve analysis indicated that, in this setting, prognostic screening does not add value over and above a treat-all approach. The potential for LBP patients to be misclassified using screening and the high incidence of nonrecovery indicate that care decisions should be made with the assumption that all patients are ‘at risk.’
Perspective
This article presents a head-to-head comparison of 3 LBP screening instruments in a secondary care setting. Early patient screening is likely to hold little clinical value in this setting and care pathways that consider all patients at risk of a poor outcome are suggested to be most appropriate.
Key words
Low back pain
prognostic screening
predictive validity
Cited by (0)
Emma L. Karran is supported by the Royal Adelaide Hospital Clinical Research Grants (2014 and 2015) and the Royal Adelaide Hospital Research foundation Dawes Scholarship (2016). G. Lorimer Moseley is supported by the National Health and Medical Research Council (NHMRC) (Principal Research Fellowship ID: 1061279). James H. McAuley is supported by the NHMRC (Project Grant ID: 008017). Adrian C. Traeger is supported by an NHMRC PhD Scholarship (ID: 1075670).
G. Lorimer Moseley has received support from workers' compensation boards in Australia, Europe, and North America, Kaiser Permanente USA, Pfizer Australia, Agile Physiotherapy USA, and Results Physiotherapy USA; he receives royalties for books on pain and rehabilitation; he receives speaker fees for lectures on pain and rehabilitation. Emma L. Karran, Adrian C. Traeger, James H. McAuley, Susan L. Hillier, and Yun-Hom Yau have no conflicts of interest to declare.