Decision-making in spine surgery is complex due to patients heterogeneity and complexity of spinal pathologies and the various surgical options applied to a given pathology. Artificial intelligence/machine learning algorithms provide an opportunity to improve patient selection, surgical planning, and outcomes. The purpose of this article is to present the experience and applications of in spine surgery at 2 large academic health care systems.

OBJECTIVE: We applied natural language processing and inference methods to extract social determinants of health (SDoH) information from clinical notes of patients with chronic low back pain (cLBP) to enhance future analyses of the associations between SDoH disparities and cLBP outcomes. MATERIALS AND METHODS: Clinical notes for patients with cLBP were annotated for 7 SDoH domains, as well as depression, anxiety, and pain scores, resulting in 626 notes with at least one annotated entity for 364 patients. We used a 2-tier taxonomy with these 10 first-level classes (domains) and 52 second-level classes. We developed and validated named entity recognition (NER) systems based on both rule-based and machine learning approaches and validated an entailment model. RESULTS: Annotators achieved a high interrater agreement (Cohens kappa of 95.3% at document level). A rule-based system (cTAKES), RoBERTa NER, and a hybrid model (combining rules and logistic regression) achieved performance of F1 = 47.1%, 84.4%, and 80.3%, respectively, for first-level classes. DISCUSSION: While the hybrid model had a lower F1 performance, it matched or outperformed RoBERTa NER model in terms of recall and had lower computational requirements. Applying an untuned RoBERTa entailment model, we detected many challenging wordings missed by NER systems. Still, the entailment model may be sensitive to hypothesis wording. CONCLUSION: This study developed a corpus of annotated clinical notes covering a broad spectrum of SDoH classes. This corpus provides a basis for training machine learning models and serves as a benchmark for predictive models for NER for SDoH and knowledge extraction from clinical texts.

Purpose

To identify independent risk factors, including the Risk Assessment and Prediction Tool (RAPT) score, associated with extended length of stay (eLOS) and non-home discharge following elective multi-level instrumented spine fusion operations for diagnosis of adult spinal deformity (ASD) and lumbar degenerative pathology.

Methods

Adults who underwent multi-level ([Formula: see text] segments) instrumented spine fusions for ASD and lumbar degenerative pathology at a single institution (2016-2021) were reviewed. Presence of a pre-operative RAPT score was used as an inclusion criterion. Excluded were patients who underwent non-elective operations, revisions, operations for trauma, malignancy, and/or infections. Outcomes were eLOS (> 7 days) and discharge location (home vs. non-home). Predictor variables included demographics, comorbidities, operative information, Surgical Invasiveness Index (SII), and RAPT score. Fisher's exact test was used for univariate analysis, and significant variables were implemented in multivariate binary logistic regression, with generation of 95% percent confidence intervals (CI), odds ratios (OR), and p-values.

Results

Included for analysis were 355 patients. Post-operatively, 36.6% (n = 130) had eLOS and 53.2% (n = 189) had a non-home discharge. Risk factors significant for a non-home discharge were older age (> 70 years), SII > 36, pre-op RAPT < 10, DMII, diagnosis of depression or anxiety, and eLOS. Risk factors significant for an eLOS were SII > 20, RAPT < 6, and an ASA score of 3.

Conclusion

The RAPT score and SII were most important significant predictors of eLOS and non-home discharges following multi-level instrumented fusions for lumbar spinal pathology and deformity. Preoperative optimization of the RAPT's individual components may provide a useful strategy for decreasing LOS and modifying discharge disposition.

Objective

Using the newly created University of California (UC) Health Data Warehouse, we present the first study to analyze antihyperglycemic treatment utilization across the five large UC academic health systems (Davis, Irvine, Los Angeles, San Diego, and San Francisco).

Research design and methods

This retrospective analysis used deidentified electronic health records (EHRs; 2014-2019) including 97,231 patients with type 2 diabetes from 1,003 UC-affiliated clinical settings. Significant differences between health systems and individual providers were identified using binomial probabilities with cohort matching.

Results

Our analysis reveals statistically different treatment utilization patterns not only between health systems but also among individual providers within health systems. We identified 21 differences among health systems and 29 differences among individual providers within these health systems, with respect to treatment intensifications within existing guidelines on top of either metformin monotherapy or dual therapy with metformin and a sulfonylurea. Next, we identified variation for medications within the same class (e.g., glipizide vs. glyburide among sulfonylureas), with 33 differences among health systems and 86 among individual providers. Finally, we identified 2 health systems and 55 individual providers who more frequently used medications with known cardioprotective benefits for patients with high cardiovascular disease risk, but also 1 health system and 8 providers who prescribed such medications less frequently for these patients.

Conclusions

Our study used cohort-matching techniques to highlight real-world variation in care between health systems and individual providers. This demonstrates the power of EHRs to quantify differences in treatment utilization, a necessary step toward standardizing precision care for large populations.

Study design

A retrospective study at a single academic institution.

Objective

The purpose of this study is to utilize machine learning to predict hospital length of stay (LOS) and discharge disposition following adult elective spine surgery, and to compare performance metrics of machine learning models to the American College of Surgeon's National Surgical Quality Improvement Program's (ACS NSQIP) prediction calculator.

Summary of background data

A total of 3678 adult patients undergoing elective spine surgery between 2014 and 2019, acquired from the electronic health record.

Methods

Patients were divided into three stratified cohorts: cervical degenerative, lumbar degenerative, and adult spinal deformity groups. Predictive variables included demographics, body mass index, surgical region, surgical invasiveness, surgical approach, and comorbidities. Regression, classification trees, and least absolute shrinkage and selection operator (LASSO) were used to build predictive models. Validation of the models was conducted on 16% of patients (N=587), using area under the receiver operator curve (AUROC), sensitivity, specificity, and correlation. Patient data were manually entered into the ACS NSQIP online risk calculator to compare performance. Outcome variables were discharge disposition (home vs. rehabilitation) and LOS (days).

Results

Of 3678 patients analyzed, 51.4% were male (n=1890) and 48.6% were female (n=1788). The average LOS was 3.66 days. In all, 78% were discharged home and 22% discharged to rehabilitation. Compared with NSQIP (Pearson R2 =0.16), the predictions of poisson regression ( R2 =0.29) and LASSO ( R2 =0.29) models were significantly more correlated with observed LOS ( P =0.025 and 0.004, respectively). Of the models generated to predict discharge location, logistic regression yielded an AUROC of 0.79, which was statistically equivalent to the AUROC of 0.75 for NSQIP ( P =0.135).

Conclusion

The predictive models developed in this study can enable accurate preoperative estimation of LOS and risk of rehabilitation discharge for adult patients undergoing elective spine surgery. The demonstrated models exhibited better performance than NSQIP for prediction of LOS and equivalent performance to NSQIP for prediction of discharge location.

Introduction

Care and outcomes for patients with chronic low back pain (cLBP) are influenced by the social risk factors that they experience. Social risk factors such as food insecurity and housing instability have detrimental effects on patient health and wellness, healthcare outcomes, and health disparities.

Objectives

This retrospective cross-sectional study examined how social risk factors identified in unstructured and structured electronic health record (EHR) data for 1,295 patients with cLBP were associated with health care utilization. We also studied the impact of social risk factors, controlling for back pain-related disability on health care utilization.

Methods

Included patients who received outpatient spine and/or physical therapy services at an urban academic medical center between 2018 and 2020. Five identified social risks were financial insecurity, housing instability, food insecurity, transportation barriers, and social isolation. Outcomes included 4 categories of health care utilization: emergency department (ED) visits/hospitalizations, imaging, outpatient specialty visits related to spine care, and physical therapy (PT) visits. Poisson regression models tested associations between the presence of identified social risks and each outcome measure.

Results

Identified social risks in 12.8% of the study population (N = 166/1,295). In multivariate models, social isolation was positively associated with imaging, specialty visits, and PT visits; housing instability was positively associated with ED visits/hospitalizations and imaging; food insecurity was positively associated with ED visits/hospitalizations and specialty visits but negatively associated with PT visits; and financial strain was positively associated with PT visits but negatively associated with ED visits/hospitalization.

Conclusion

These associations were seen above and beyond other factors used as markers of socioeconomic marginalization, including neighborhood-level social determinants of health, race/ethnicity, and insurance type. Identifying and intervening on social risk factors that patients with cLBP experience may improve outcomes and be cost-saving.

Background

Tools, such as the STarTBack Screening Tool (SBT), have been developed to identify risks of progressing to chronic disability in low back pain (LBP) patients in the primary care population. However, less is known about predictors of change in function after treatment in the specialty care population.

Objective

We pursued a retrospective observational cohort study involving LBP patients seen in a multidisciplinary specialty clinic to assess which features can predict change in function at follow-up.

Methods

The SBT was administered at initial visit, and a variety of patient characteristics were available in the chart including the presence of chronic overlapping pain conditions (COPCs). Patient Reported Outcomes Measurement Information System-10 (PROMIS-10) global physical health (PH) and global mental health (MH) were measured at baseline and at pragmatic time points during follow-up. Linear regression was used to estimate adjusted associations between available features and changes in PROMIS scores.

Results

241 patients were followed for a mean of 17.0 ± 7.5 months. Mean baseline pain was 6.7 (SD 2.1), PROMIS-10 global MH score was 44.8 (SD 9.3), and PH score was 39.4 (SD 8.6). 29.7% were low-risk on the SBT, 41.8% were medium-risk, and 28.5% were high-risk. Mean change in MH and PH scores from baseline to the follow-up questionnaire were 0.86 (SD 8.11) and 2.39 (SD 7.52), respectively. Compared to low-risk patients, high-risk patients had a mean 4.35 points greater improvement in their MH score (p= 0.004) and a mean 3.54 points greater improvement in PH score (p= 0.006). Fewer COPCs also predicted greater improvement in MH and PH.

Conclusions

SBT and the presence of COPC, which can be assessed at initial presentation to a specialty clinic, can predict change in PROMIS following treatment. Effort is needed to identify other factors that can help predict change in function after treatment in the specialty care setting.

Chronic low back pain (LBP) is a leading cause of disability and opioid prescriptions worldwide, representing a significant medical and socioeconomic problem. Clinical heterogeneity of LBP limits accurate diagnosis and precise treatment planning, culminating in poor patient outcomes. A current priority of LBP research is the development of objective, multidimensional assessment tools that subgroup LBP patients based on neurobiological pain mechanisms, to facilitate matching patients with the optimal therapies. Using unsupervised machine learning on full body biomechanics, including kinematics, dynamics, and muscle forces, captured with a marker-less depth camera, this study identified a forward-leaning sit-to-stand strategy (STS) as a discriminating movement biomarker for LBP subjects. A forward-leaning STS strategy, as opposed to a vertical rise strategy seen in the control participants, is less efficient and results in increased spinal loads. Inefficient STS with the subsequent higher spinal loading may be a biomarker of poor motor control in LBP patients as well as a potential source of the ongoing symptomology.

There is a great and growing need to ascertain what exactly is the state of a patient, in terms of disease progression, actual care practices, pathology, adverse events, and much more, beyond the paucity of data available in structured medical record data. Ascertaining these harder-to-reach data elements is now critical for the accurate phenotyping of complex traits, detection of adverse outcomes, efficacy of off-label drug use, and longitudinal patient surveillance. Clinical notes often contain the most detailed and relevant digital information about individual patients, the nuances of their diseases, the treatment strategies selected by physicians, and the resulting outcomes. However, notes remain largely unused for research because they contain Protected Health Information (PHI), which is synonymous with individually identifying data. Previous clinical note de-identification approaches have been rigid and still too inaccurate to see any substantial real-world use, primarily because they have been trained with too small medical text corpora. To build a new de-identification tool, we created the largest manually annotated clinical note corpus for PHI and develop a customizable open-source de-identification software called Philter ("Protected Health Information filter"). Here we describe the design and evaluation of Philter, and show how it offers substantial real-world improvements over prior methods.