Large language models to identify social determinants of health in electronic health records.

Social determinants of health (SDoH) play a critical role in patient outcomes, yet their documentation is often missing or incomplete in the structured data of electronic health records (EHRs). Large language models (LLMs) could enable high-throughput extraction of SDoH from the EHR to support research and clinical care. However, class imbalance and data limitations present challenges for this sparsely documented yet critical information. Here, we investigated the optimal methods for using LLMs to extract six SDoH categories from narrative text in the EHR: employment, housing, transportation, parental status, relationship, and social support. The best-performing models were fine-tuned Flan-T5 XL for any SDoH mentions (macro-F1 0.71), and Flan-T5 XXL for adverse SDoH mentions (macro-F1 0.70). Adding LLM-generated synthetic data to training varied across models and architecture, but improved the performance of smaller Flan-T5 models (delta F1 + 0.12 to +0.23). Our best-fine-tuned models outperformed zero- and few-shot performance of ChatGPT-family models in the zero- and few-shot setting, except GPT4 with 10-shot prompting for adverse SDoH. Fine-tuned models were less likely than ChatGPT to change their prediction when race/ethnicity and gender descriptors were added to the text, suggesting less algorithmic bias (p < 0.05). Our models identified 93.8% of patients with adverse SDoH, while ICD-10 codes captured 2.0%. These results demonstrate the potential of LLMs in improving real-world evidence on SDoH and assisting in identifying patients who could benefit from resource support.

Development and Validation of an Automated Image-Based Deep Learning Platform for Sarcopenia Assessment in Head and Neck Cancer.

Importance: Sarcopenia is an established prognostic factor in patients with head and neck squamous cell carcinoma (HNSCC); the quantification of sarcopenia assessed by imaging is typically achieved through the skeletal muscle index (SMI), which can be derived from cervical skeletal muscle segmentation and cross-sectional area. However, manual muscle segmentation is labor intensive, prone to interobserver variability, and impractical for large-scale clinical use. Objective: To develop and externally validate a fully automated image-based deep learning platform for cervical vertebral muscle segmentation and SMI calculation and evaluate associations with survival and treatment toxicity outcomes. Design, Setting, and Participants: For this prognostic study, a model development data set was curated from publicly available and deidentified data from patients with HNSCC treated at MD Anderson Cancer Center between January 1, 2003, and December 31, 2013. A total of 899 patients undergoing primary radiation for HNSCC with abdominal computed tomography scans and complete clinical information were selected. An external validation data set was retrospectively collected from patients undergoing primary radiation therapy between January 1, 1996, and December 31, 2013, at Brigham and Women's Hospital. The data analysis was performed between May 1, 2022, and March 31, 2023. Exposure: C3 vertebral skeletal muscle segmentation during radiation therapy for HNSCC. Main Outcomes and Measures: Overall survival and treatment toxicity outcomes of HNSCC. Results: The total patient cohort comprised 899 patients with HNSCC (median [range] age, 58 [24-90] years; 140 female [15.6%] and 755 male [84.0%]). Dice similarity coefficients for the validation set (n = 96) and internal test set (n = 48) were 0.90 (95% CI, 0.90-0.91) and 0.90 (95% CI, 0.89-0.91), respectively, with a mean 96.2% acceptable rate between 2 reviewers on external clinical testing (n = 377). Estimated cross-sectional area and SMI values were associated with manually annotated values (Pearson r = 0.99; P < .001) across data sets. On multivariable Cox proportional hazards regression, SMI-derived sarcopenia was associated with worse overall survival (hazard ratio, 2.05; 95% CI, 1.04-4.04; P = .04) and longer feeding tube duration (median [range], 162 [6-1477] vs 134 [15-1255] days; hazard ratio, 0.66; 95% CI, 0.48-0.89; P = .006) than no sarcopenia. Conclusions and Relevance: This prognostic study's findings show external validation of a fully automated deep learning pipeline to accurately measure sarcopenia in HNSCC and an association with important disease outcomes. The pipeline could enable the integration of sarcopenia assessment into clinical decision making for individuals with HNSCC.

F-18 FDG perfusion PET: intraprocedural assessment of the liver tumor ablation margin.

PURPOSE: To evaluate 18F-fluorodeoxyglucose (FDG) perfusion PET during FDG PET/CT-guided liver tumor microwave ablation procedures for assessing the ablation margin and correlating minimum margin measurements with local progression. METHODS: This IRB-approved, HIPAA-compliant study included 20 adult patients (11 M, 9 F; mean age 65) undergoing FDG PET/CT-guided liver microwave ablation to treat 31 FDG-avid tumors. Intraprocedural FDG perfusion PET was performed to assess the ablation margin. Intraprocedural decisions regarding overlapping ablations were recorded. Two readers retrospectively interpreted intraprocedural perfusion PET and postprocedural contrast-enhanced MRI. Assessability of the ablation margin and minimum margin measurements were recorded. Imaging follow-up for local progression ranged from 30 to 574 days (mean 310). Regression modeling of minimum margin measurements was performed. Hazard ratios were calculated to correlate an ablation margin threshold of 5 mm with outcomes. RESULTS: Intraprocedural perfusion PET prompted additional overlapping ablations of two tumors, neither of which progressed. Incomplete ablation or local progression occurred in 8/31 (26%) tumors. With repeat ablation, secondary efficacy was 26 (84%) of 31. Both study readers deemed ablation margins fully assessable more often using perfusion PET than MRI (OR 69.7; CI 6.0, 806.6; p = 0.001). Minimum ablation margins ≥ 5 mm on perfusion PET correlated with a low risk of incomplete ablation/local progression by both study readers (HR 0.08 and 0.02, p < 0.001). CONCLUSION: Intraprocedural FDG perfusion PET consistently enabled complete liver tumor microwave ablation margin assessments, and the perfusion PET minimum ablation margin measurements correlated well with local outcomes. Clinical trial registration clinicaltrials.gov (NCT02018107).

Population-based estimates of survival among elderly patients with brain metastases.

BACKGROUND: Prognostic estimates for patients with brain metastases (BM) stem from younger, healthier patients enrolled in clinical trials or databases from academic centers. We characterized population-level prognosis in elderly patients with BM. METHODS: Using Surveillance, Epidemiology, and End Results (SEER)-Medicare data, we identified 9882 patients ≥65 years old with BM secondary to lung, breast, skin, kidney, esophageal, colorectal, and ovarian primaries between 2014 and 2016. Survival was assessed by primary site and evaluated with Cox regression. RESULTS: In total, 2765 versus 7117 patients were diagnosed with BM at primary cancer diagnosis (synchronous BM, median survival = 2.9 mo) versus thereafter (metachronous BM, median survival = 3.4 mo), respectively. Median survival for all primary sites was ≤4 months, except ovarian cancer (7.5 mo). Patients with non-small-cell lung cancer (NSCLC) receiving epidermal growth factor receptor (EGFR)- or anaplastic lymphoma kinase (ALK)-based therapy for synchronous BM displayed notably better median survival at 12.5 and 20.1 months, respectively, versus 2.8 months exhibited by other patients with NSCLC; survival estimates in melanoma patients based on receipt of BRAF/MEK therapy versus not were 6.7 and 2.8 months, respectively. On multivariable regression, older age, greater comorbidity, and type of managing hospital were associated with poorer survival; female sex, higher median household income, and use of brain-directed stereotactic radiation, neurosurgical resection, or systemic therapy (versus brain-directed non-stereotactic radiation) were associated with improved survival (all P < 0.05). CONCLUSIONS: Elderly patients with BM have a poorer prognosis than suggested by prior algorithms. If prognosis is driven by systemic and not intracranial disease, brain-directed therapy with potential for significant toxicity should be utilized cautiously.

Racial disparities in supportive medication use among older patients with brain metastases: a population-based analysis.

BACKGROUND: Brain metastases (BM) cause symptoms that supportive medications can alleviate. We assessed whether racial disparities exist in supportive medication utilization after BM diagnosis. METHODS: Medicare-enrolled patients linked with the Surveillance, Epidemiology, and End Results program (SEER) who had diagnoses of BM between 2007 and 2016 were identified. Fourteen supportive medication classes were studied: non-opioid analgesics, opioids, anti-emetics, anti-epileptics, headache-targeting medications, steroids, cognitive aids, antidepressants, anxiolytics, antidelirium/antipsychotic agents, muscle relaxants, psychostimulants, sleep aids, and appetite stimulants. Drug administration ≤30 days following BM diagnosis was compared by race using multivariable logistic regression. RESULTS: Among 17,957 patients, headache aids, antidepressants, and anxiolytics were prescribed less frequently to African Americans (odds ratio [95% CI] = 0.81 [0.73-0.90], P < 0.001; OR = 0.68 [0.57-0.80], P < 0.001; and OR = 0.68 [0.56-0.82], P < 0.001, respectively), Hispanics (OR = 0.83 [0.73-0.94], P = 0.004 OR = 0.78 [0.64-0.97], P = 0.02; and OR = 0.63 [0.49-0.81], P < 0.001, respectively), and Asians (OR = 0.81 [0.72-0.92], P = 0.001, OR = 0.67 [0.53-0.85], P = 0.001, and OR = 0.62 [0.48-0.80], P < 0.001, respectively) compared with non-Hispanic Whites. African Americans also received fewer anti-emetics (OR = 0.75 [0.68-0.83], P < 0.001), steroids (OR = 0.84 [0.76-0.93], P < 0.001), psychostimulants (OR = 0.14 [0.03-0.59], P = 0.007), sleep aids (OR = 0.71 [0.61-0.83], P < 0.001), and appetite stimulants (OR = 0.85 [0.77-0.94], P = 0.002) than Whites. Hispanic patients less frequently received antidelirium/antipsychotic drugs (OR = 0.57 [0.38-0.86], P = 0.008), sleep aids (OR = 0.78 [0.64-0.94, P = 0.01), and appetite stimulants (OR = 0.87 [0.76-0.99], P = 0.04). Asian patients received fewer opioids (OR = 0.86 [0.75-0.99], P = 0.04), anti-emetics (OR = 0.83 [0.73-0.94], P = 0.004), anti-epileptics (OR = 0.83 [0.71-0.97], P = 0.02), steroids (OR = 0.81 [0.72-0.92], P = 0.001), muscle relaxants (OR = 0.60 [0.41-0.89], P = 0.01), and appetite stimulants (OR = 0.87 [0.76-0.99], P = 0.03). No medication class was prescribed significantly less frequently to Whites. CONCLUSIONS: Disparities in supportive medication prescription for non-White/Hispanic groups with BM exist; improved provider communication and engagement with at-risk patients is needed. KEY POINTS: 1. Patients with BM commonly experience neurologic symptoms.2. Supportive medications improve quality of life among patients with BM.3. Non-White patients with BM receive fewer supportive medications than White patients.

Funding Support and Principal Investigator Leadership of Oncology Clinical Trials Using Radiation Therapy.

PURPOSE: Sources of funding and principal investigator (PI) leadership for clinical trials using radiation therapy (RT) are not well characterized but are important mediators of innovation, particularly because funding for trials from the National Institutes of Health (NIH) has decreased and industry funding has increased. We sought to determine characteristics of trials using RT that are associated with industry funding, NIH funding, and radiation oncologist (RO) PI leadership. METHODS AND MATERIALS: www.ClinicalTrials.gov was queried for all open, interventional trials that administered RT. Logistic regression was used to identify associations between trial characteristics, receipt of funding type (NIH, industry, or other), and PI leadership. RESULTS: The authors identified 1469 oncology trials, of which 41% were based in the United States, 56% were based internationally, and 3% were based in the United States and internationally. Of these, 22% were RT monotherapy, 53% were bimodality (40% RT + drug, 13% RT + surgery), and 24% were trimodality. Although ROs led 60% of all trials, industry-sponsored trials were significantly less likely to have RO PIs (35% RO vs 65% non-RO PI; adjusted odds ratio [aOR], 0.45; 95% confidence interval [CI], 0.28-0.73), to fund trials that did not incorporate drug therapy (aOR, 0.19; 95% CI, 0.10-0.35), or to fund phase III trials (aOR, 0.25; 95% CI, 0.11-0.60) because industry-sponsored trials favored smaller phase I trials. NIH-funded trials were not associated with PI type and, although not statistically significant, favored larger phase III trials (unadjusted OR, 2.06; 95% CI, 0.99-4.29). ROs were less likely to lead trials incorporating drug therapy (aOR, 0.30; 95% CI, 0.22-0.41). CONCLUSIONS: ROs are less likely than other specialties to lead trials that use RT in combination with drug therapy or surgery and more likely to lead trials supported by nonindustry, non-NIH funding. This suggests a need for ROs to lead multimodality trials and to consider opportunities to interact with industry. As NIH resources decrease, alternative funding is needed to support innovation, particularly in in RT-alone trials.

An Assessment of the Quality Oncology Practice Initiative: Lessons Learned From a Detailed Assessment of a Well-Established Profession-Based Performance Measurement Program.

BACKGROUND: National performance measurement programs used to improve quality and/or accountability are prevalent. Professional society-based programs largely assess practice or provider performance. We tested the limitations of the Quality Oncology Practice Initiative (QOPI), a well-established program. METHODS: We investigated the potential limitations of the QOPI limited sampling strategy and end-of-life (EOL) metrics through a retrospective review of all decedents meeting the QOPI eligibility criteria at a large three-site community oncology practice. Relative precision for each EOL metric was measured and simulated scenarios modeled via log-normal distributions were compared. RESULTS: A total of 246 deaths were identified; only 14% of decedents were included in the QOPI limited sample. Important differences in relative precision between samples were evident. Chemotherapy administered at the EOL was 2.8 times greater among the full cohort compared to the sample. CONCLUSION: The limited sampling strategy demonstrated the lack of precision explained by sampling variability confounded in smaller sample sizes. Our analyses demonstrated that practices reporting the same metric may stem from very different underlying distributions, which limits the ability to understand complex practice patterns and improve care. These findings may be of significance to other performance measurement programs that utilize similar testing strategies.

Healthier Standards for School Meals and Snacks: Impact on School Food Revenues and Lunch Participation Rates.

INTRODUCTION: In 2012, the updated U.S. Department of Agriculture school meals standards and a competitive food law similar to the fully implemented version of the national Smart Snack standards went into effect in Massachusetts. This study evaluated the impact of these updated school meal standards and Massachusetts' comprehensive competitive food standards on school food revenues and school lunch participation. METHODS: Revenue and participation data from 11 Massachusetts school districts were collected from 2011 to 2014 and analyzed in 2015 using multilevel modeling. The association between the change in compliance with the competitive food standards and revenues/participation was assessed using linear regression. RESULTS: Schools experienced declines in school food revenues of $15.40/student in Year 1 from baseline (p=0.05), due to competitive food revenue losses. In schools with 3 years of data, overall revenues rebounded by the second year post-implementation. Additionally, by Year 2, school lunch participation increased by 15% (p=0.0006) among children eligible for reduced-price meals. Better competitive food compliance was inversely associated with school food revenues in the first year only; an absolute change in compliance by 10% was associated with a $9.78/student decrease in food revenues over the entire school year (p=0.04). No association was seen between the change in compliance and school meal participation. CONCLUSIONS: Schools experienced initial revenue losses after implementation of the standards, yet longer-term school food revenues were not impacted and school meal participation increased among children eligible for reduced-price meals. Weakening the school meal or competitive food guidelines based on revenue concerns appears unwarranted.

Comparison of performance of various tumor response criteria in assessment of sunitinib activity in advanced gastrointestinal stromal tumors.

PURPOSE: To compare the performance of various tumor response criteria (TRC) in the assessment of patients with advanced gastrointestinal stromal tumor (GIST) treated with sunitinib after failure of imatinib. METHODS: Sixty-two participants with advanced GIST in two clinical trials received oral sunitinib after prior failure of imatinib (median duration 24 weeks; interquartile range 14-56) and were followed with contrast-enhanced computed tomography at baseline and thereafter at median intervals of 6 weeks (IRQ 6-9). Tumor response was prospectively determined using Response Evaluation Criteria in Solid Tumors (RECIST) 1.0, and retrospectively reassessed for comparison using RECIST 1.1, Choi criteria, and modified Choi (mChoi) criteria using the original target lesions. For mChoi criteria, progressive disease was defined as 20% increase in sum of the longest dimension, similar to RECIST 1.1. Clinical benefit rate (CBR; complete response, partial response, or stable disease ≥12 weeks) and progression-free survival were compared between various TRCs using kappa statistics. RESULTS: While partial response as the best response was more frequent by Choi and mChoi criteria (50% each) than RECIST 1.1 (15%) and RECIST 1.0 (13%), CBR was similar between various TRCs (overall CBR 60%-77%, 77%-94% agreement between all TRC pairs). Time to best response was shorter for Choi and mChoi criteria (median 11 weeks each) compared to RECIST 1.1 and RECIST 1.0 (median 25 and 24 weeks, respectively). PFS was similar for RECIST 1.1, RECIST 1.0, and mChoi (median 35 weeks each), and shortest for Choi criteria (median 23 weeks). CONCLUSIONS: CBR was similar among the various TRCs, although Choi criteria led to earlier determination of disease progression. Therefore, RECIST 1.1 and mChoi criteria may be preferred for response assessment in patients with advanced GIST.

Prospective assessment of deep inspiration breath-hold using 3-dimensional surface tracking for irradiation of left-sided breast cancer.

PURPOSE: Deep inspiration breath hold (DIBH) is used to decrease cardiac irradiation during radiation therapy (RT) for breast cancer. The patients most likely to benefit and the impact on treatment time remain largely unknown. We sought to identify predictors for the use of DIBH and to quantify differences in dosimetry and treatment time using a prospective registry. METHODS AND MATERIALS: A total of 150 patients with left breast cancer were enrolled. All patients were simulated with both free breathing (FB) and DIBH. RT was delivered by either modality. Alternate scans were planned with use of deformable registration to include identical RT volumes. DIBH patients were monitored by a real-time surface tracking system, AlignRT (Vision RT, Ltd, London, United Kingdom). Baseline characteristics and treatment times were compared by Fisher exact test and Wilcoxon rank sum test. Dosimetric endpoints were analyzed by Wilcoxon signed rank test, and linear regression identified predictors for change in mean heart dose (∆MHD). RESULTS: We treated 38 patients with FB and 110 with DIBH. FB patients were older, more likely to have heart and lung disease, and less likely to receive chemotherapy or immediate reconstruction (all P < .05). Treatment times were not significantly different, but DIBH patients had greater variability in times (P = .0002). Of 146 evaluable patients, DIBH resulted in >20 cGy improvement in MHD in 107 patients but a >20 cGy increase in MHD in 14. Both MHD and lung V20 were significantly lower in DIBH than in paired FB plans. On multivariate analysis, younger age (4.18 cGy per year; P < .0001), higher body mass index (6.06 cGy/kg/m(2); P = .0018), and greater change in lung volumes (130 cGy/L; P = .003) were associated with greater ∆MHD. CONCLUSIONS: DIBH improves cardiac dosimetry without significantly impacting treatment time in most patients. Greater inspiratory lung volumes augment this benefit. Because the improvement with DIBH was not uniform, patients should be scanned with both FB and DIBH.

Feasibility of using algorithm-based clinical decision support for symptom assessment and management in lung cancer.

CONTEXT: Distressing symptoms interfere with the quality of life in patients with lung cancer. Algorithm-based clinical decision support (CDS) to improve evidence-based management of isolated symptoms seems promising, but no reports yet address multiple symptoms. OBJECTIVES: This study examined the feasibility of CDS for a Symptom Assessment and Management Intervention targeting common symptoms in patients with lung cancer (SAMI-L) in ambulatory oncology. The study objectives were to evaluate completion and delivery rates of the SAMI-L report and clinician adherence to the algorithm-based recommendations. METHODS: Patients completed a web-based symptom assessment and SAMI-L created tailored recommendations for symptom management. Completion of assessments and delivery of reports were recorded. Medical record review assessed clinician adherence to recommendations. Feasibility was defined as 75% or higher report completion and delivery rates and 80% or higher clinician adherence to recommendations. Descriptive statistics and generalized estimating equations were used for data analyses. RESULTS: Symptom assessment completion was 84% (95% CI=81-87%). Delivery of completed reports was 90% (95% CI=86-93%). Depression (36%), pain (30%), and fatigue (18%) occurred most frequently, followed by anxiety (11%) and dyspnea (6%). On average, overall recommendation adherence was 57% (95% CI=52-62%) and was not dependent on the number of recommendations (P=0.45). Adherence was higher for anxiety (66%; 95% CI=55-77%), depression (64%; 95% CI=56-71%), pain (62%; 95% CI=52-72%), and dyspnea (51%; 95% CI=38-64%) than for fatigue (38%; 95% CI=28-47%). CONCLUSION: The CDS systems, such as SAMI-L, have the potential to fill a gap in promoting evidence-based care.

Few changes in food security and dietary intake from short-term participation in the Supplemental Nutrition Assistance Program among low-income Massachusetts adults.

OBJECTIVE: To examine whether short-term participation in the Supplemental Nutrition Assistance Program (SNAP) affects food security and dietary quality among low-income adults recruited from a Massachusetts-wide emergency food hotline. METHODS: A 3-month, longitudinal study was conducted among 107 adults recruited at the time of SNAP application assistance. Outcomes included household food security (10-item US Department of Agriculture Food Security Survey Module), dietary intake (eg, grains, fruit) and diet quality (modified Alternate Healthy Eating Index). Data were analyzed using paired t tests and multivariable linear regression. RESULTS: Supplemental Nutrition Assistance Program participation was not associated with improved household food security over 3 months (P = .25). Compared with non-participants, SNAP participants increased refined grain intake by 1.1 serving/d (P = .02), from baseline to follow-up. No associations were observed with other foods, nutrients, or dietary quality. CONCLUSION AND IMPLICATIONS: Policies that simultaneously improve household food security and dietary quality should be implemented to support the health of low-income Americans participating in this crucial program.

Dietary intake and dietary quality of low-income adults in the Supplemental Nutrition Assistance Program.

BACKGROUND: The Supplemental Nutrition Assistance Program (SNAP) aims to alleviate hunger among its beneficiaries by providing benefits to purchase nutritious foods. OBJECTIVE: We conducted a comprehensive dietary analysis of low-income adults and examined differences in dietary intake between SNAP participants and nonparticipants. DESIGN: The study population comprised 3835 nonelderly adults with a household income ≤130% of the federal poverty level from the 1999-2008 NHANES. The National Cancer Institute method was used to estimate the distributions of usual intake for dietary outcomes. Relative differences in dietary intake by SNAP participation were estimated with adjustment for sociodemographic characteristics and household food security. RESULTS: Few low-income adults consumed recommended amounts of whole grains, fruit, vegetables, fish, and nuts/seeds/legumes. Conversely, many low-income adults exceeded recommended limits for processed meats, sweets, and bakery desserts and sugar-sweetened beverages. Approximately 13-22% of low-income adults did not meet any food and nutrient guidelines; virtually no adults met all of the guidelines. Compared with nonparticipants, SNAP participants consumed 39% fewer whole grains (95% CI: -57%, -15%), 44% more 100% fruit juice (95% CI: 0%, 107%), 56% more potatoes (95% CI: 18%, 106%), 46% more red meat (95% CI: 4%, 106%), and, in women, 61% more sugar-sweetened beverages (95% CI: 3%, 152%). SNAP participants also had lower dietary quality scores than did nonparticipants, as measured by a modified Alternate Healthy Eating Index. CONCLUSION: Although the diets of all low-income adults need major improvement, SNAP participants in particular had lower-quality diets than did income-eligible nonparticipants.

Technological advancements and error rates in radiation therapy delivery.

PURPOSE: Technological advances in radiation therapy (RT) delivery have the potential to reduce errors via increased automation and built-in quality assurance (QA) safeguards, yet may also introduce new types of errors. Intensity-modulated RT (IMRT) is an increasingly used technology that is more technically complex than three-dimensional (3D)-conformal RT and conventional RT. We determined the rate of reported errors in RT delivery among IMRT and 3D/conventional RT treatments and characterized the errors associated with the respective techniques to improve existing QA processes. METHODS AND MATERIALS: All errors in external beam RT delivery were prospectively recorded via a nonpunitive error-reporting system at Brigham & Women's Hospital/Dana Farber Cancer Institute. Errors are defined as any unplanned deviation from the intended RT treatment and are reviewed during monthly departmental quality improvement meetings. We analyzed all reported errors since the routine use of IMRT in our department, from January 2004 to July 2009. Fisher's exact test was used to determine the association between treatment technique (IMRT vs. 3D/conventional) and specific error types. Effect estimates were computed using logistic regression. RESULTS: There were 155 errors in RT delivery among 241,546 fractions (0.06%), and none were clinically significant. IMRT was commonly associated with errors in machine parameters (nine of 19 errors) and data entry and interpretation (six of 19 errors). IMRT was associated with a lower rate of reported errors compared with 3D/conventional RT (0.03% vs. 0.07%, p = 0.001) and specifically fewer accessory errors (odds ratio, 0.11; 95% confidence interval, 0.01-0.78) and setup errors (odds ratio, 0.24; 95% confidence interval, 0.08-0.79). CONCLUSIONS: The rate of errors in RT delivery is low. The types of errors differ significantly between IMRT and 3D/conventional RT, suggesting that QA processes must be uniquely adapted for each technique. There was a lower error rate with IMRT compared with 3D/conventional RT, highlighting the need for sustained vigilance against errors common to more traditional treatment techniques.

A simulation study of quantitative risk assessment for bivariate continuous outcomes.

The neurotoxic effects of chemical agents are often investigated in controlled studies on rodents, with binary and continuous multiple endpoints routinely collected. One goal is to conduct quantitative risk assessment to determine safe dose levels. Yu and Catalano (2005) describe a method for quantitative risk assessment for bivariate continuous outcomes by extending a univariate method of percentile regression. The model is likelihood based and allows for separate dose-response models for each outcome while accounting for the bivariate correlation. The approach to benchmark dose (BMD) estimation is analogous to that for quantal data without having to specify arbitrary cutoff values. In this article, we evaluate the behavior of the BMD relative to background rates, sample size, level of bivariate correlation, dose-response trend, and distributional assumptions. Using simulations, we explore the effects of these factors on the resulting BMD and BMDL distributions. In addition, we illustrate our method with data from a neurotoxicity study of parathion exposure in rats.

Quantitative risk assessment for multivariate continuous outcomes with application to neurotoxicology: the bivariate case.

The neurotoxic effects of chemical agents are often investigated in controlled studies on rodents, with multiple binary and continuous endpoints routinely collected. One goal is to conduct quantitative risk assessment to determine safe dose levels. Such studies face two major challenges for continuous outcomes. First, characterizing risk and defining a benchmark dose are difficult. Usually associated with an adverse binary event, risk is clearly definable in quantal settings as presence or absence of an event; finding a similar probability scale for continuous outcomes is less clear. Often, an adverse event is defined for continuous outcomes as any value below a specified cutoff level in a distribution assumed normal or log normal. Second, while continuous outcomes are traditionally analyzed separately for such studies, recent literature advocates also using multiple outcomes to assess risk. We propose a method for modeling and quantitative risk assessment for bivariate continuous outcomes that address both difficulties by extending existing percentile regression methods. The model is likelihood based; it allows separate dose-response models for each outcome while accounting for the bivariate correlation and overall characterization of risk. The approach to estimation of a benchmark dose is analogous to that for quantal data without the need to specify arbitrary cutoff values. We illustrate our methods with data from a neurotoxicity study of triethyl tin exposure in rats.