Steven Z George’s research while affiliated with North Carolina Clinical Research and other places

Background Outcomes from Total Joint Arthroplasty (TJA) are variable but generally favorable. However, the literature is lacking regarding direct comparisons of important outcomes across TJA sites. Such comparisons are of paramount importance to informing future bundled care reform and patient optimization. Thus, we compared the rates of adverse events, filled prescriptions, and costs at 90 days and 365 days after TJA for knee, hip, and shoulder patients. Methods We conducted a retrospective cohort study of multi-payor claims data with patients ( n = 2416) who underwent hip ( n = 909), knee ( n = 1250), or shoulder ( n = 257) TJA within an academic health system. Univariable and multivariable logistic regression models were used to assess the association between the TJA surgical site and adverse events (i.e., medical and surgical complications) and prescriptions filled. Univariable and multivariable gamma regression models were used to assess the association between the TJA surgical site and total cost and surgical episode cost. Results In all regression models, the hip location was used as the reference group. There were no differences in the adjusted odds of medical complications between the TJA surgical sites after adjusting for confounders at 90 days or 365 days. For surgical complications, the adjusted odds were 2.66 times higher in the knee ( P < 0.001) and 4.48 times higher in the shoulder ( P < 0.001) at 90 days. At 365 days, the odds were 2.54 times higher in the knee ( P < 0.001) and 4.10 times higher in the shoulder ( P < 0.001). There was an increase in the adjusted odds of antiepileptic and NSAIDS being filled in knee and shoulder patients compared to hip patients at 31–90 days (both P < 0.001). At 0–365 days, knee patients had increased adjusted odds of filled antibiotic ( P = 0.032), antiepileptic ( P = 0.001), and opioid ( P = 0.005) prescriptions compared to hip patients, while shoulder patients only increased odds of antiepileptic ( P = 0.028). Lastly, in adjusted models, both the knee and shoulder had a significant increase in total health system costs, with a 9% and 14% increase in cost, respectively ( P < 0.01). Conclusion Patients undergoing TKA and TSA may have an increased risk for surgical complications and longer-term opioid prescriptions (TKA only) compared to those undergoing THA. Collectively, these results can inform future population-based approaches to managing osteoarthritis care pathways or reimbursement policies for TJA across multiple joint sites.

Purpose Measures of pain sensitivity have potential relevance for patient care. We previously identified a subgroup of people at risk for ongoing pain characterized by genetic AND psychological factors. Here, we report planned secondary analyses examining the effect of personalized interventions on pain sensitivity outcomes. Patients and Methods Two hundred and sixty-one healthy individuals with the COMT SNP rs6269 AA genotype and Pain Catastrophizing Scale scores of 5 or higher received exercise-induced muscle injury, followed by a randomly assigned treatment: (1) general education and placebo; (2) personalized psychological intervention and placebo; (3) general education and propranolol; or (4) personalized psychological intervention and propranolol. Pain sensitivity outcomes (pressure pain thresholds (PPT), suprathreshold heat rating, temporal summation, and conditioned pain modulation efficiency) were compared using a mixed effect model to examine difference among groups, adjusted for age, sex and race. Results No main effects for group assignment were noted (p > 0.05 for all), when considered as 4 groups or 2 collapsed groups (ie propranolol vs placebo or personalized psychologic vs general education). Interaction terms were then entered into our models in an exploratory fashion. For PPT outcomes interactions were noted for, sex and time, and race and time (p<0.015). For temporal summation outcomes, interactions were noted for sex and group and race and group (p < 0.015). Conclusion Results indicated no statistically reliable changes in pain sensitivity when considering matched vs unmatched treatment groups. Caution is needed in this interpretation given that the trial was not powered to specifically identify these differences. Exploratory analysis of interactions among ethnic/racial and gender identities by treatment, however, showed the potential for differential effects for specific pain sensitivity measures. Significant interactions across modalities suggest analysis of higher order interactions/intersectionality could be of great interest for testing efficacy of personalized interventions in future trials.

Introduction Low back pain (LBP) is a leading global factor in disability among older adults. Movement-evoked pain (MEP) is potentially an important mediator in the disability pathway but is predominantly tested in the laboratory. Objectives We aimed to explore MEP in the natural environment (“daily” MEP) and its correlation with laboratory MEP, along with potential psychological and immunological influences. Method Thirty-five older adults with persistent LBP attended a single laboratory session. Pain catastrophizing, pain-related fear of movement, and pain self-efficacy were measured by questionnaire. Resting inflammation and inflammatory reactivity to painful movement were evaluated using serum interleukin-6, tissue necrosis factor alpha, and C-reactive protein (CRP). Laboratory MEP was defined by aggregate pain intensity with a movement provocation test. Daily MEP was measured for the next 7 days using ecological momentary assessment. Results Laboratory MEP was strongly correlated with daily MEP ( ρ = 0.780, P = <0.001). C-reactive protein (Hedges [ g ] = 0.266) and interleukin-6 (g = 0.433) demonstrated small to moderate reactivity to painful movement. After controlling for age and multimorbidity, pain catastrophizing and pain self-efficacy explained 24% to 37% variance in laboratory and daily MEP. Resting inflammatory markers were not associated with MEP; however, C-reactive protein reactivity to painful movement explained 19% to 25% variance in laboratory and daily MEP. Conclusion Preliminary indication is that laboratory and daily MEP may be proxy measures for one another, and that MEP is influenced by psychological and immunological factors. Future studies will aim to (1) validate findings among older adults with persistent LBP and (2) for clinical phenotyping, clarify complex relationships among psychological and immunological factors with disability pathway components like MEP.

Background Embedded pragmatic clinical trials (ePCTs) assess interventions in real-world settings. Best practices for recruiting clinical sites for ePCTs are unknown, especially for sites that aren't known to the study team or familiar with clinical research. We describe the site recruitment process for AIM-Back, an ePCT of two nonpharmacologic pathways for low back pain within the Veterans Health Administration (VA). Methods During the planning phase of the AIM-Back trial, we aimed to recruit 18–20 sites. Eligible sites required provider capacity, administrative support, and geographic separation to avoid contamination. Our three-step approach involved: (1) lead (VA personnel) identification through existing VA contacts, data repositories of VA clinicians, and promotional outreach at events and listservs; (2) lead engagement via tailored communications emphasizing participation benefits; and (3) virtual meetings with administrators and clinicians. Results We identified 184 leads across 53 VA healthcare systems. Leads from 40 systems responded to outreach, and recruitment meetings were conducted with 23 systems involving primary care, physical therapy, research staff, and leadership. We met our recruitment goal, securing participation agreements from 19 sites, with a median timeline from outreach to participation agreement of 3.7 months. Common reasons for non-participation included infrastructure and resource constraints, resistance to new clinical programs, and competing programs. Conclusion AIM-Back's recruitment highlights ePCT site recruitment complexities for trials engaging new clinical research sites. Our innovative three-step recruitment approach provides an example for similarly designed trials. Future ePCTs should consider comprehensive recruitment strategies to ensure clinician buy-in, study feasibility, and broaden existing networks for completing ePCTs.

Background Delivering evidence-based interventions remains challenging, particularly for complex conditions like chronic musculoskeletal pain. Non-pharmacologic treatments are recommended for many pain conditions, but implementing these can be difficult due to their complexity and resource demands. Pragmatic trials, especially embedded designs, provide a method to see how interventions are being implemented and adapted in real-world settings throughout the trial process. This study explored how intervention delivery complexity and adaptations differ between non-pharmacologic pain trials and non-pain trials to provide guidance on future treatment delivery and implementation. Methods From July to October 2023, an online survey was distributed to members of three NIH Trial Collaboratories to assess intervention delivery complexity and adaptations during their pragmatic trials. Participants rated their trial's intervention delivery complexity using a 7-item tool and reported any adaptations to intervention delivery throughout the trial process. Data analysis compared complexity and adaptations between the two trial types to explore differences and relationships between intervention delivery complexity and adaptations. Results We analyzed 12 pain and 12 non-pain trials and found that intervention delivery complexity was not discernibly different between the two trial types, however, pain trials did have a slightly higher average intervention delivery complexity, overall. Pain trials also had more adaptations in the workflow domain compared to non-pain trials, while adaptations across other domains were similar between the two types. Workflow emerged as the most challenging domain for adaptation among all trials. Conclusion Intervention delivery complexity may be higher for pragmatic trials that are investigating non-pharmacologic pain interventions versus non-pain trials, but only in very specific areas.

Background A number of efforts have been made to tailor behavioral healthcare treatments to the variable needs of patients with low back pain (LBP). The most common approach involves the STarT Back Screening Tool (SBST) to triage the need for psychologically informed care, which explores concerns about pain and addresses unhelpful beliefs, attitudes, and behaviors. Such beliefs that pain always signifies injury or tissue damage and that exercise should be avoided have been implied as psychosocial mediators of chronic pain and can impede recovery. The ability of physical therapy interventions guided by baseline stratification for risk of persistent LBP or related functional limitations to improve unhelpful pain beliefs has not been well assessed. Because treatments are aimed at addressing these beliefs, understanding a bit more about the nature of beliefs about pain (for example, attitudes and knowledge) might help us understand how to better tailor this care or even our risk-stratification approaches for future treatment of patients with LBP. Questions/purposes (1) Did patients assigned to receive risk-stratified care score higher on an assessment of pain science knowledge? (2) Did patients assigned to receive risk-stratified care have fewer unhelpful attitudes related to pain? (3) Did patients assigned to receive risk-stratified care have less pain-associated psychological distress? (4) Regardless of intervention received, is baseline SBST risk category (low, medium, or high) associated with changes in attitudes and knowledge about pain? Methods This is a secondary analysis of short-term changes in pain beliefs following the 6-week treatment phase of a randomized controlled trial that examined the effectiveness of a risk-stratified physical therapy intervention on pain-related disability at 1 year. Between April 2017 and February 2020, a total of 290 patients in the Military Health System seeking primary care for LBP were enrolled in a trial comparing a behavioral-based intervention to usual care. The intervention involved psychologically informed physical therapy using cognitive behavioral principles and included tailored education, graded exercise, and graded exposure. Individuals assigned to usual care followed treatment plans set forth by their primary care provider. Thirty-one patients were removed from Optimal Screening for Prediction of Referral and Outcome Yellow Flag (OSPRO-YF) tool analyses due to missing assessments at 6 weeks (n = 15 intervention; n = 16 usual care). This resulted in 89% (259 of 290) of participants included for secondary analysis, with no difference in baseline demographic characteristics between groups. The usual-care group comprised 50% of the total study group (129 of 259), with a mean age of 34 ± 9 years; 67% (87 of 129) were men. The risk-stratified care group comprised 50% (130 of 259) of the total study group, with a mean ± SD age of 35 ± 8 years; 64% (83 of 130) were men. Six additional individuals were removed from Survey of Pain Attitudes harm scale (SOPA-h) and revised Neurophysiology of Pain Questionnaire (rNPQ) analyses for missing baseline data (n = 1 intervention) and 6-week data (n = 2 intervention; n = 3 usual care). The rNPQ captured current pain science knowledge, the SOPA-h examined patient attitudes about pain (the extent of beliefs that pain leads to damage and that movement is harmful), and the OSPRO-YF assessed patients for yellow flag clinical markers of pain-related psychological distress across 11 constructs within domains of negative mood, fear avoidance, and positive affect/coping indicative of elevated vulnerability and decreased resilience. Outcomes were assessed at baseline and 6 weeks, and data were analyzed per protocol. We assessed between-group differences at 6 weeks using linear mixed-effects models of pain attitudes and knowledge and related distress, controlling for age, gender, and baseline pain. Regardless of treatment group, we also analyzed differences in rNPQ and SOPA-h scores at 6 weeks based on SBST risk category (low versus medium or high) using generalized linear (Gaussian) regression models. Results Risk-stratified treatment was associated with improvements in pain knowledge (rNPQ mean difference 6% [95% confidence interval (CI) 1% to 11%]; p = 0.01) and a reduction in indicators of pain-associated psychological distress (OSPRO-YF mean difference -1 [95% CI -2 to 0]; p = 0.01) at 6 weeks compared with usual care. There was no difference between groups for SOPA-h score at 6 weeks (mean difference -0.2 [95% CI -0.3 to 0.0]; p = 0.09). Patients with medium- or high-risk scores on the SBST, regardless of intervention, improved slightly more on SOPA-h (β = -0.31; p < 0.01) but not rNPQ (β = 0.02; p = 0.95) than those scoring low risk. Conclusion Patients receiving risk-stratified care showed small improvements in pain knowledge and reductions in pain-related psychological distress at 6 weeks, immediately after intervention, compared with usual care. Implementation of this risk-stratified care approach for LBP was able to change patients’ perceptions about pain and reduce some of their psychological distress beyond what was achieved by usual care in this setting. As these factors are believed to favorably mediate treatment outcomes, future studies should investigate whether these improvements persist over the long term, determine how they influence clinical outcomes, and explore alternatives for risk stratification and treatment to elicit greater improvements. Level of Evidence Level III, therapeutic study.

Background: Clinical practice guidelines (CPGs) are developed to synthesize evidence into recommendations for clinical practice. Minimal evidence exists on the evaluation practice of physical therapists in the treatment of patients with neck pain. Purpose: We sought to describe (1) the extent to which clinicians perform the Neck Pain CPG-recommended examination measures and (2) the percentage of patients properly classified. Methods: We retrospectively analyzed the electronic health records of 397 patients with neck pain at an ambulatory care setting in an academic medical center. The frequency of physical therapists’ evaluation measures, subjective findings, positive examination results, and the percentage of patients properly classified into impairment-based categories (IBCs) were recorded. Descriptive statistics and χ ² tests were used to assess patient demographics and compare classification accuracy across IBCs. Results: Of the 397 patients, 56% were classified into an IBC. The most common IBC was neck pain with mobility deficits (24%), followed by neck pain with radiating pain (17%), neck pain with movement coordination impairments (NPMCIs) (8%), and neck pain with headache (6%). Neck pain with movement coordination impairment had the lowest percentage of proper classifications. Classification accuracy was highest when subjective and objective findings were combined and varied between IBCs. Conclusion: Our findings suggest that physical therapists evaluating patients with neck pain may have increased classification accuracy when subjective and objective findings are considered. Decreased classification accuracy was demonstrated in the NPMCI category, highlighting opportunities for further education and research.