Penalized Spline-Involved Tree-based (PenSIT) Learning for estimating an optimal dynamic treatment regime using observational data.

Dynamic treatment regimes are a set of time-adaptive decision rules that can be used to personalize treatment across multiple stages of care. Grounded in causal inference methods, dynamic treatment regimes identify variables that differentiate the treatment effect and may be used to tailor treatments across individuals based on the patient's own characteristics - thereby representing an important step toward personalized medicine. In this manuscript we introduce Penalized Spline-Involved Tree-based Learning, which seeks to improve upon existing tree-based approaches to estimating an optimal dynamic treatment regime. Instead of using weights determined from the estimated propensity scores, which may result in unstable estimates when weights are highly variable, we predict missing counterfactual outcomes using regression models that incorporate a penalized spline of the propensity score and other covariates predictive of the outcome. We further develop a novel purity measure applied within a decision tree framework to produce a flexible yet interpretable method for estimating an optimal multi-stage multi-treatment dynamic treatment regime. In simulation experiments we demonstrate good performance of Penalized Spline-Involved Tree-based Learning relative to competing methods and, in particular, we show that Penalized Spline-Involved Tree-based Learning may be advantageous when the sample size is small and/or when the level of confounding of the outcome is high. We apply Penalized Spline-Involved Tree-based Learning to the retrospectively-collected Medical Information Mart for Intensive Care dataset to identify variables that may be used to tailor early fluid resuscitation strategies in septic patients.

Modifiable Factors Associated With Chronic Pain 1 Year After Operative Management of Distal Radius Fractures: A Secondary Analysis of a Randomized Clinical Trial.

Importance: Despite appropriate treatment, many patients who sustain distal radius fractures (DRFs) report persistent wrist pain. Chronic musculoskeletal pain is among the leading health problems in the elderly population associated with significant personal and societal burden. Objective: To identify modifiable preoperative factors that are significantly associated with developing chronic pain. Design, Setting, and Participants: This is a secondary analysis of the Wrist and Radius Injury Surgical Trial (WRIST), a randomized multicenter clinical trial of 24 study sites in the United States, Canada, and Singapore that enrolled patients from April 10, 2012, to December 31, 2016. Adults older than 60 years who sustained closed extra-articular DRFs, were treated operatively, and completed 12-month Michigan Hand Outcomes Questionnaires (MHQs) were included in this study. Analysis was conducted from September to December 2019. Interventions: Volar locking plate internal fixation, external fixation, or percutaneous pinning. Main Outcomes and Measures: 12-month MHQ pain domain score. Inverse probability weighted logistic regression was used to identify factors associated with of chronic pain. Results: A total of 146 patients with DRF who were treated operatively and had 12-month MHQ scores met inclusion criteria. The mean (SD) patient age was 68.9 (7.2) years, 128 (87.6%) were women, and 93 (63.7%) were retired. Chronic pain was present in 87 patients (59.6%) and absent in 59 patients (40.4%) at 1-year follow-up. A 1-week delay in surgery was associated with more than triple the odds of developing chronic pain (odds ratio [OR], 3.65; 95% CI, 1.48-9.00), and each 10-point increase in preoperative pain was associated with a 17% increase in the odds of experiencing chronic pain (OR, 1.17; 95% CI, 1.02-1.34). Internal fixation was associated with decreased odds of developing chronic pain compared with the other 2 procedures (OR, 0.29; 95% CI, 0.10-0.90). Conclusions and Relevance: In this study, preoperative pain, time to surgery, and procedure type were modifiable factors associated with chronic pain 1 year after DRF treated with surgery. Adequate pain control in patients with acute DRFs even before definitive surgical management and earlier fixation for patients requiring surgery may decrease the risk of developing chronic pain. Internal fixation may decrease the risk of chronic pain after DRF surgery, compared with percutaneous pinning or external fixation. Trial Registration: ClinicalTrials.gov Identifier: NCT01589692.

Trigger Finger Treatment: Identifying Predictors of Nonadherence and Cost.

BACKGROUND: Evidence-based practices in medicine are linked with a higher quality of care and lower health care cost. For trigger finger, identifying patient factors associated with nonadherence to evidence-based practices will aid physicians in treatment decisions. The objectives were to (1) determine patient factors associated with treatment nonadherence, (2) examine the success rates of steroid injections, and (3) evaluate the economic consequences of nonadherence to treatment recommendations. METHODS: The authors used data from the Clinformatics DataMart database from 2010 to 2017 to conduct a population-based analysis of patients with single-digit trigger finger. The authors calculated rates of steroid injection success and examined associations between injection success and patient factors using chi-square tests. In addition, the authors analyzed differences in the cost to the insurer, the cost to the patient, and total cost. RESULTS: A total of 29,722 patients were included in this analysis. Injection success rates were similar for diabetic (72 percent) and nondiabetic patients (73 percent), women (73 percent), and men (73 percent). Nonetheless, diabetics (OR, 1.4; 95 percent CI, 1.4 to 1.5; p < 0.001) and women (OR, 1.2; 95 percent CI, 1.1 to 1.2; p < 0.001) were significantly more likely to receive nonadherent treatment. In total, $23 million (U.S. dollars) were spent on nonadherent trigger finger care. CONCLUSIONS: Diabetics and women have increased odds of having surgery without a prior steroid injection, despite similar success rates of steroid injections compared to nondiabetics and men. Because performing surgical release before any steroid injections may represent a higher cost treatment option, providers should provide steroid injections before surgery for all patients regardless of diabetes status or sex to minimize overtreatment. CLINICAL QUESTION/LEVEL OF EVIDENCE: Risk, III.

Assessment of Tree-Based Statistical Learning to Estimate Optimal Personalized Treatment Decision Rules for Traumatic Finger Amputations.

Importance: Optimal treatment for traumatic finger amputation is unknown to date. Objective: To use statistical learning methods to estimate evidence-based treatment assignment rules to enhance long-term functional and patient-reported outcomes in patients after traumatic amputation of fingers distal to the metacarpophalangeal joint. Design, Setting, and Participants: This decision analytical model used data from a retrospective cohort study of 338 consenting adult patients who underwent revision amputation or replantation at 19 centers in the United States and Asia from August 1, 2016, to April 12, 2018. Of those, data on 185 patients were included in the primary analysis. Exposures: Treatment with revision amputation or replantation. Main Outcomes and Measures: Outcome measures were hand strength, dexterity, hand-related quality of life, and pain. A tree-based statistical learning method was used to derive clinical decision rules for treatment of traumatic finger amputation. Results: Among 185 study participants (mean [SD] age, 45 [16] years; 156 [84%] male), the median number of fingers amputated per patient was 1 (range, 1-5); 115 amputations (62%) were distal to the proximal interphalangeal joint, and 110 (60%) affected the nondominant hand. On the basis of the tree-based statistical learning estimates, to maximize hand dexterity or to minimize patient-reported pain, replantation was found to be the best strategy. To maximize hand strength, revision amputation was the best strategy for patients with a single-finger amputation but replantation was preferred for all other injury patterns. To maximize patient-reported quality of life, revision amputation was the best approach for patients with dominant hand injuries, and replantation was the best strategy for patients with nondominant hand injuries. Conclusions and Relevance: The findings suggest that the approach to treating traumatic finger amputations varies based on the patient's injury characteristics and functional needs.

Local Control and Toxicity of Multilevel Spine Stereotactic Body Radiotherapy.

BACKGROUND: Spine stereotactic body radiotherapy (sSBRT) is commonly limited to 1 or 2 vertebral levels given a paucity of efficacy and toxicity data when more than 2 levels are treated. OBJECTIVE: To prove our hypothesis that multilevel sSBRT could provide similar rates of local control (LC) (primary endpoint) and toxicity as single-level treatment using the same clinical target, planning target, and planning organ-at-risk volumes. METHODS: We analyzed consecutive cases of sSBRT treated from 2013 to 2017. Time-to-event outcomes for single-level and multilevel cases were compared using mixed effect Cox models and differences in toxicity rates were evaluated using linear mixed effect models. All models incorporate a patient-level random intercept to account for any within-patient correlation across cases. RESULTS: There were 101 single-level and 84 multilevel sSBRT cases (2-7 continuous vertebral levels). One-year LC was 95% vs 85%, respectively. After adjusting for baseline covariates, dose delivered, and accounting for within-patient correlation, there was no significant difference in time to local failure (hazard ratio, HR 1.79 [0.59-5.4]; P = .30). Pain improved in 83.5% of the 139 initially symptomatic tumors. There were no significant differences in grade 2+ acute or late toxicities between single-level and multilevel sSBRT. CONCLUSION: With rigorous patient immobilization, quality assurance, and image guidance, multilevel sSBRT provides high rates of LC, similar to single-level treatment, without need for larger planning volume margins. Efforts to improve prognostication and case selection for multilevel sSBRT are warranted to ensure that the benefits of improved LC over palliative radiation are justified.

Patient satisfaction after treatment of distal radial fractures in older adults.

The Wrist and Radius Injury Surgical Trial (WRIST) collaboration is the largest clinical trial ever conducted in hand surgery. We applied data from this study to examine the relationship between functional outcomes and patient satisfaction after treatment of distal radial fractures. Patients aged 60 years and older with isolated distal radial fractures were enrolled at 24 health systems. Grip strength and the arc of wrist motion were measured after treatment. The Michigan Hand Outcomes Questionnaire was used to measure patient satisfaction. Receiver operating characteristic curves were created using patient satisfaction as the reference standard and each functional measure as a predictor. At 12 months after treatment, mean grip strength was 82% of normal and mean arc of motion was 88% of normal. The optimal cut-off points to distinguish satisfaction from dissatisfaction occurred when patients recovered 59% of hand strength and 79% of wrist motion. Continuing therapy to increase functional gains beyond this point unnecessarily utilizes healthcare resources without additional patient-reported gains. Level of evidence: IV.

Assessment of Surgeon Variation in Adherence to Evidence-Based Recommendations for Treatment of Trigger Finger.

Importance: Stenosing tenosynovitis (trigger finger) affects approximately 2% of the population. Given the prevalence of trigger finger and rising health care costs, adherence to the cost-effective and evidence-based treatment algorithm will permit better outcomes and allocation of resources. Objectives: To examine treatment patterns for trigger finger and to determine surgeon-level and patient-level factors that influence adherence to evidence-based treatment. Design, Setting, and Participants: This retrospective population-based cohort study examined deidentified claims for treatment of trigger finger from a national insurance provider using the Clinformatics Data Mart database. Patients were included if they were 18 years or older and treated from January 1, 2002, through December 31, 2016 (excluding a washout period from July 1, 2008, until June 30, 2010), with a new diagnosis of single-digit trigger finger. Data were analyzed from December 21, 2018, through April 28, 2019. Exposures: Cost-effective and evidence-based research published in July 2009 for the treatment of trigger finger. Main Outcomes and Measures: After excluding the 1-year washout period on either side of July 1, 2009, adherence to the recommended treatment algorithm of 2 corticosteroid injections before surgical release of trigger finger was compared with practice before publication of research supporting this cost-effective and evidence-based approach. Results: In this analysis of 83 667 patients with trigger finger, 52 698 (63.0%) were women, and 20 045 (24.0%) had type 1 or 2 diabetes. Mean (SD) age was 61 (13) years. From 2002 to 2016, an overall increasing trend in adherence to the cost-effective and evidence-based approach to treatment was noted, with no significant increase in adherence in the postpublication era (67.5% vs 73.3%; P = .27). Substantial variation in adherence was observed at the surgeon level (intraclass correlation, 33%). Plastic surgeons had no change in adherence over time compared with orthopedic surgeons (odds ratio [OR], 1.00; 95% CI, 0.98-1.02; P = .90), whereas general surgeons had increased adherence (OR, 1.04; 95% CI, 1.02-1.06; P < .001). Higher-volume surgeons were also more adherent to these evidence-based recommendations (OR, 1.59; 95% CI, 1.53-1.65; P < .001). Conclusions and Relevance: This study found substantial surgeon-level variation in adherence to evidence-based treatment of trigger finger. Surgeon specialty and volume were associated with differences in adherence. Efforts to understand surgeon barriers to implementation, regardless of physician specialty, appear to be necessary, and better implementation strategies may permit increased uptake of evidence-based treatment of trigger finger.

Radiation-Induced Skin Changes after Postmastectomy Radiation Therapy: A Pilot Study on Indicators for Timing of Delayed Breast Reconstruction.

BACKGROUND: Delayed autologous breast reconstruction is commonly recommended in patients requiring postmastectomy radiation. This study examines gross and histologic changes in the breast skin of patients who have undergone postmastectomy radiation to help determine when radiation-induced skin changes begin to stabilize. METHODS: A prospective pilot study was conducted on eight patients with invasive breast cancer who required mastectomy and radiotherapy. At the time of mastectomy and 2, 4, 6, 8, and 12 months after completion of radiotherapy, a punch biopsy was taken from the radiated mastectomy skin of each patient. Serial standardized photographs were taken before and after radiotherapy to evaluate the degree of hyperpigmentation and graded by three blinded plastic surgeons. Skin biopsies were processed for histologic assessment of inflammation, elastin organization, and vascularity. RESULTS: Grading of patient photographs revealed an increase in hyperpigmentation after radiotherapy compared with baseline with a gradual improvement over time. SMAD3 immunostaining demonstrated a trend toward an increase in inflammation over 12 months. The elastin distribution within samples showed an increase in fiber disorganization, thickening, and clumping, with no improvement throughout the study period. The average number of vessels per high powered field decreased steadily through the duration of the study. CONCLUSION: Histologic changes in dermal inflammation, elastin organization, and vascularity do not appear to correspond with the gradual improvement of hyperpigmentation, resulting from postmastectomy radiation. These histologic changes persist beyond the 12 month observation period and will require clinical correlation to determine the potential impact on postoperative outcomes.

Effect of Aromatase Inhibitor Therapy on Sleep and Activity Patterns in Early-stage Breast Cancer.

INTRODUCTION: Adherence to aromatase inhibitor (AI) therapy is poor, often because of treatment-emergent side effects, including musculoskeletal symptoms, fatigue, and insomnia. In the present analysis, we examined the sleep patterns and daytime function both objectively using actigraphy and subjectively using validated questionnaires in women initiating AI therapy. PATIENTS AND METHODS: Postmenopausal women with stage 0-III hormone receptor-positive breast cancer who were initiating AI therapy were eligible. The patients wore actigraphy devices for 10 consecutive days and completed questionnaires at baseline before the initiation of AI and after 3 months of AI therapy. Associations between the baseline demographics and symptoms, changes in patient-reported outcomes and actigraphy measures from baseline to 3 months of AI therapy and discontinuation of AI therapy were examined using sign tests, logistic regression models, Spearman's correlation, and linear mixed models. RESULTS: Forty-two patients (86%) completed the baseline assessments and 23 patients (47%) completed both the baseline and the 3-month assessments. Objectively measured daytime function as measured by total daytime activity decreased during the 3 months after starting AI (232,566 activity count vs. 204,205 activity count; P = .023), and the decrease was more evident in women with higher baseline physical function. Reduced daytime activity correlated with increased fatigue (ρ = -0.49; P = .017). CONCLUSION: Daytime function decreased after initiation of AI therapy and correlated moderately with increased fatigue, although no association was identified with changes in pain or sleep quality. Additional studies are required to understand why function is reduced, which could have implications for interventions to improve patient tolerance of, and persistence with, AI therapy.

Variable aromatase inhibitor plasma concentrations do not correlate with circulating estrogen concentrations in post-menopausal breast cancer patients.

PURPOSE: The aromatase inhibitors (AI) exemestane (EXE), letrozole (LET), and anastrozole suppress estrogen biosynthesis, and are effective treatments for estrogen receptor (ER)-positive breast cancer. Prior work suggests that anastrozole blood concentrations are associated with the magnitude of estrogen suppression. The objective of this study was to determine whether the magnitude of estrogen suppression, as determined by plasma estradiol (E2) concentrations, in EXE or LET treated patients is associated with plasma AI concentrations. METHODS: Five hundred post-menopausal women with ER-positive breast cancer were enrolled in the prospective Exemestane and Letrozole Pharmacogenetic (ELPh) Study conducted by the COnsortium on BReast cancer phArmacogomics (COBRA) and randomly assigned to either drug. Estrogen concentrations were measured at baseline and after 3 months of AI treatment and drug concentrations were measured after 1 or 3 months. EXE or LET concentrations were compared with 3-month E2 concentration or the change from baseline to 3 months using several complementary statistical procedures. RESULTS: Four-hundred patients with on-treatment E2 and AI concentrations were evaluable (EXE n = 200, LET n = 200). Thirty (7.6%) patients (EXE n = 13, LET n = 17) had 3-month E2 concentrations above the lower limit of quantification (LLOQ) (median: 4.75; range: 1.42-63.8 pg/mL). EXE and LET concentrations were not associated with on-treatment E2 concentrations or changes in E2 concentrations from baseline (all p > 0.05). CONCLUSIONS: Steady-state plasma AI concentrations do not explain variability in E2 suppression in post-menopausal women receiving EXE or LET therapy, in contrast with prior evidence in anastrozole treated patients.