This is the authors' abstract. We don't add key points for this paper.
Dear Editor, Low-back pain (LBP) is 1 of the leading causes of years lived with disability,1 with chronic low-back pain (CLBP) representing most of the costs and disability among those with LBP.2 However, the lack of a consistent definition for CLBP makes it difficult to compare different studies.3 Given that the term CLBP is used as a prognostic indicator, this study assessed which patient-reported definition of CLBP provides the best indication of CLBP and high-impact back pain 6 months later for adults who report LBP at baseline. We studied adults who responded “Yes, I currently have this condition” to the question “Do you currently have back pain?” at baseline in 2 online panels: (1) Amazon Mechanical Turk (MTurk), a convenience panel of workers who participate in the Amazon online MTurk marketplace;4 and (2) KnowledgePanel (KP), the oldest and largest probability-based online panel in the United States.5 We identified those who met 1 or more of 4 definitions of CLBP at baseline:3 3-month duration definition: Response of at least 3 months or more to “how long has LBP been an ongoing problem for you?” Research Task Force (RTF)6definition: Responses of at least 3-months or more to “How long has LBP been an ongoing problem for you?” and at least half the days in the past 6 months or more to “How often has your LBP been an ongoing problem?” Provider identified: Response of “Yes” to “Has a health provider told you that your back pain is chronic?” Individual identified: Response of “Yes” to “Do you think your back pain is chronic?” High-impact pain was defined in 2 ways: a score of grade 2 or higher on the Graded Chronic Pain Scale (GCPS), or responding “most days” or “every day” to “Over the past 3 months, how often did pain limit your life or work activities?”7–9 For each CLBP definition at baseline, we calculated its sensitivity (ability to identify all those with condition), specificity (ability to identify those without condition), Youden index (balance of sensitivity and specificity assuming equal weight for each),10 positive predictive value (PPV, ability to identify a true positive among screener positives), negative predictive value (NPV, ability to identify a true negative among screener negatives), and accuracy (percentage of correct predictions) in predicting LBP and high-impact pain 6 months later. We included 827 individuals in the MTurk sample and 1235 in the KP sample (see Table 1). The MTurk sample was younger, more likely to be non-Hispanic, more educated, and more likely to have high-impact pain at baseline than the KP sample. Over 90% in each sample met at least 1 definition of CLBP at baseline. In both samples, the CLBP definition most often met was 3-month duration (∼90%), and the least common was provider-identified (∼33%). While meeting at least 1 definition of CLBP was not significantly different between samples, the proportion of those meeting the RTF (P < .001) and individual-identified (P = .003) definitions of chronic pain were higher among MTurk participants while meeting the provider-identified (P = .030) definition was higher among KP participants. High-impact pain in MTurk (KP) was 53% (44%) at baseline and 47% (50%) 6-months later. Characteristics of study participants. Significance test results (*** P < .001,** P < .01,* P < .05). HS = High school; KP = data collected from Ipsos’s KnowledgePanel; MTurk = data collected using Amazon’s Mechanical Turk; RTF = NIH Pain Consortium Research Task Force. Measures of the 6-months predictive power of the different CLBP definitions at baseline were similar between the MTurk and KP datasets (see Table 2). At least 1 CLBP definition was a better-than-chance indicator of future back pain according to most of these measures except the PPV for future activity limitation high-impact pain where no definition did better than 50%. Each definition, other than patient-identified, was best at identifying future back pain according to several measures. The most sensitive predictor was the 3-month duration definition, while the provider-identified definition was the most specific. The provider-identified definition had the highest PPV for 6-month LBP and high-impact pain, whereas the 3-month definition had the highest NPV for 6-month LBP, and the RTF definition had the highest NPV for 6-month high-impact pain. The RTF definition also had the highest Youden index for 6-month LBP and GCPS high-impact pain, whereas the provider-identified definition had the highest Youden index for activity limitation high-impact pain. The 3-month definition was the most accurate at identifying 6-month LBP but was the least accurate for predicting high-impact pain at 6-months. Prediction of low back pain and high-impact pain at 6 months by different definitions of chronic low back pain at baseline FN = False negative; FP = False positive; GCPS = Graded Chronic Pain Scale; KP = data collected from Ipsos’s Knowledge Panel; MTurk = data collected using Amazon’s Mechanical Turk; NPV = Negative predictive value; PPV = Positive predictive value; RTF = NIH Pain Consortium Research Task Force; TN = True negative; TP = True positive. Bold indicates the largest value in each column. Different definitions of chronic low back pain are better at indicating future pain according to a variety of measures. This study provides guidance to clinicians and researchers as to the implications of using each CLBP definition. In choosing a definition of CLBP to use clinicians and researchers must balance the need for overall accuracy and sensitivity and specificity. If identifying all those likely to have future back pain is most important—that is, there is a substantial cost to missing some of them (Type II error), then the focus should be on maximizing sensitivity. If it is most important to minimize those categorized as having CLBP who won’t have future pain—that is, there is a substantial cost to someone being told they have CLBP when they don’t (Type I error), the focus should be on specificity. If these costs are equal, the Youden index should be used. If it is most important to be accurate in your indicator of future pain among those categorized as having CLBP, then the focus should be on PPV, and if it is most important to be accurate in your indicator of no future pain among those categorized as not having CLBP, then the focus should be on NPV. If accuracy in both directions is equally important, then use overall accuracy to choose your CLBP definition. Finally, when choosing a definition clinicians and researchers also must decide whether the goal is to identify those with future LBP or future high-impact pain. Our study has limitations. Data were self-report and from online samples. The MTurk sample (a convenience panel) may not be representative of the general population. However, the KP sample was designed to be representative of the US population. The similarity seen in the results in the 2 samples provides support for the findings. Funding for this manuscript was provided by the National Institutes of Health (1R01AT010402-01A1) to Patricia M Herman and Ron D Hays. Conflicts of interest: Nabeel Qureshi—no conflict. Ron D Hays—no conflict. Patricia M Herman—no conflict.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Qureshi et al. (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: