Finding Medicine's Moneyball: How Lessons From Major League Baseball Can Advance Assessment in Precision Education.

ABSTRACT: Precision education (PE) leverages longitudinal data and analytics to tailor educational interventions to improve patient, learner, and system-level outcomes. At present, few programs in medical education can accomplish this goal as they must develop new data streams transformed by analytics to drive trainee learning and program improvement. Other professions, such as Major League Baseball (MLB), have already developed extremely sophisticated approaches to gathering large volumes of precise data points to inform assessment of individual performance.In this perspective, the authors argue that medical education-whose entry into precision assessment is fairly nascent-can look to MLB to learn the possibilities and pitfalls of precision assessment strategies. They describe 3 epochs of player assessment in MLB: observation, analytics (sabermetrics), and technology (Statcast). The longest tenured approach, observation, relies on scouting and expert opinion. Sabermetrics brought new approaches to analyzing existing data in a way that better predicted which players would help the team win. Statcast created precise, granular data about highly attributable elements of player performance while helping to account for nonplayer factors that confound assessment such as weather, ballpark dimensions, and the performance of other players. Medical education is progressing through similar epochs marked by workplace-based assessment, learning analytics, and novel measurement technologies. The authors explore how medical education can leverage intersectional concepts of MLB player and medical trainee assessment to inform present and future directions of PE.

Association between duration of smoking abstinence before non-small-cell lung cancer diagnosis and survival: a retrospective, pooled analysis of cohort studies.

BACKGROUND: The association between duration of smoking abstinence before non-small-cell lung cancer (NSCLC) diagnosis and subsequent survival can influence public health messaging delivered in lung-cancer screening. We aimed to assess whether the duration of smoking abstinence before diagnosis of NSCLC is associated with improved survival. METHODS: In this retrospective, pooled analysis of cohort studies, we used 26 cohorts participating in Clinical Outcomes Studies of the International Lung Cancer Consortium (COS-ILCCO) at 23 hospitals. 16 (62%) were from North America, six (23%) were from Europe, three (12%) were from Asia, and one (4%) was from South America. Patients enrolled were diagnosed between June 1, 1983, and Dec 31, 2019. Eligible patients had smoking data before NSCLC diagnosis, epidemiological data at diagnosis (obtained largely from patient questionnaires), and clinical information (retrieved from medical records). Kaplan-Meier curves and multivariable Cox models (ie, adjusted hazard ratios [aHRs]) were generated with individual, harmonised patient data from the consortium database. We estimated overall survival for all causes, measured in years from diagnosis date until the date of the last follow-up or death due to any cause and NSCLC-specific survival. FINDINGS: Of 42 087 patients with NSCLC in the COS-ILCCO database, 21 893 (52·0%) of whom were male and 20 194 (48·0%) of whom were female, we excluded 4474 (10·6%) with missing data. Compared with current smokers (15 036 [40·0%] of 37 613), patients with 1-3 years of smoking abstinence before NSCLC diagnosis (2890 [7·7%]) had an overall survival aHR of 0·92 (95% CI 0·87-0·97), patients with 3-5 years of smoking abstinence (1114 [3·0%]) had an overall survival aHR of 0·90 (0·83-0·97), and patients with more than 5 years of smoking abstinence (10 841 [28·8%]) had an overall survival aHR of 0·90 (0·87-0·93). Improved NSCLC-specific survival was observed in 4301 (44%) of 9727 patients who had quit cigarette smoking and was significant at abstinence durations of more than 5 years (aHR 0·87, 95% CI 0·81-0·93). Results were consistent across age, sex, histology, and disease-stage distributions. INTERPRETATION: In this large, pooled analysis of cohort studies across Asia, Europe, North America, and South America, overall survival was improved in patients with NSCLC whose duration of smoking abstinence before diagnosis was as short as 1 year. These findings suggest that quitting smoking can improve overall survival, even if NSCLC is diagnosed at a later lung-cancer screening visit. These findings also support the implementation of public health smoking cessation strategies at any time. FUNDING: The Alan B Brown Chair, The Posluns Family Fund, The Lusi Wong Fund, and the Princess Margaret Cancer Foundation.

Using Recurrent Neural Networks to Extract High-Quality Information From Lung Cancer Screening Computerized Tomography Reports for Inter-Radiologist Audit and Feedback Quality Improvement.

PURPOSE: Lung cancer screening programs generate a high volume of low-dose computed tomography (LDCT) reports that contain valuable information, typically in a free-text format. High-performance named-entity recognition (NER) models can extract relevant information from these reports automatically for inter-radiologist quality control. METHODS: Using LDCT report data from a longitudinal lung cancer screening program (8,305 reports; 3,124 participants; 2006-2019), we trained a rule-based model and two bidirectional long short-term memory (Bi-LSTM) NER neural network models to detect clinically relevant information from LDCT reports. Model performance was tested using F1 scores and compared with a published open-source radiology NER model (Stanza) in an independent evaluation set of 150 reports. The top performing model was applied to a data set of 6,948 reports for an inter-radiologist quality control assessment. RESULTS: The best performing model, a Bi-LSTM NER recurrent neural network model, had an overall F1 score of 0.950, which outperformed Stanza (F1 score = 0.872) and a rule-based NER model (F1 score = 0.809). Recall (sensitivity) for the best Bi-LSTM model ranged from 0.916 to 0.991 for different entity types; precision (positive predictive value) ranged from 0.892 to 0.997. Test performance remained stable across time periods. There was an average of a 2.86-fold difference in the number of identified entities between the most and the least detailed radiologists. CONCLUSION: We built an open-source Bi-LSTM NER model that outperformed other open-source or rule-based radiology NER models. This model can efficiently extract clinically relevant information from lung cancer screening computerized tomography reports with high accuracy, enabling efficient audit and feedback to improve quality of patient care.

Cigarette smoking cessation, duration of smoking abstinence, and head and neck squamous cell carcinoma prognosis.

BACKGROUND: Tobacco use is a major risk factor for developing head and neck squamous cell carcinoma (HNSCC). However, the prognostic associations with smoking cessation are limited. The authors assessed whether smoking cessation and increased duration of abstinence were associated with improved overall (OS) and HNSCC-specific survival. METHODS: Clinicodemographic and smoking data from patients with HNSCC at Princess Margaret Cancer Center (2006-2019) were prospectively collected. Multivariable Cox and Fine and Gray competing-risk models were used to assess the impact of smoking cessation and duration of abstinence on overall mortality and HNSCC-specific/noncancer mortality, respectively. RESULTS: Among 2482 patients who had HNSCC, former smokers (adjusted hazard ratio [aHR], 0.71; 95% CI, 0.58-0.87; p = .001; N = 841) had a reduced risk of overall mortality compared with current smokers (N = 931). Compared with current smokers, former smokers who quit >10 years before diagnosis (long-term abstinence; n = 615) had the most improved OS (aHR, 0.72; 95% CI, 0.56-0.93; p = .001). The 5-year actuarial rates of HNSCC-specific and noncancer deaths were 16.8% and 9.4%, respectively. Former smokers (aHR, 0.71; 95% CI, 0.54-0.95; p = .019) had reduced HNSCC-specific mortality compared with current smokers, but there was no difference in noncancer mortality. Abstinence for >10 years was associated with decreased HNSCC-specific death compared with current smoking (aHR, 0.64; 95% CI, 0.46-0.91; p = .012). Smoking cessation with a longer duration of quitting was significantly associated with reduced overall and HNSCC-specific mortality in patients who received primary radiation. CONCLUSIONS: Smoking cessation before the time of diagnosis reduced overall mortality and cancer-specific mortality among patients with HNSCC, but no difference was observed in noncancer mortality. Long-term abstinence (>10 pack-years) had a significant OS and HNSCC-specific survival benefit.

Incidence of kidney stones in pregnancy and associations with adverse obstetrical outcomes: a systematic review and meta-analysis of 4.7 million pregnancies.

BACKGROUND: Renal stones are a common cause of non-obstetrical abdominal pain in pregnant women. Though the management of renal stones in pregnancy is challenging, it remains unclear how the incidence of kidney stones may affect the course of pregnancy and delivery. OBJECTIVE: To determine the incidence of renal stones in pregnancy and its impact on adverse obstetrical outcomes. DATA SOURCES: We conducted a systematic literature search of three databases: Ovid MEDLINE, Ovid EMBASE, and EBSCO CINAHL Plus. After the selection of articles, an additional hand-search of their citations was completed to maximize sensitivity. Databases were examined from the last four decades (19 March 1970) up to the search date (19 March 2020). STUDY ELIGIBILITY CRITERIA: Articles were excluded if they were not relevant to kidney stones or did not report outcomes related to pregnancy. Case reports, animal studies, and cadaveric studies were excluded. Conference abstracts, gray literature, and unpublished data were not eligible. STUDY APPRAISAL AND SYNTHESIS METHODS: All screening, extraction, and synthesis were completed in duplicate with two independent reviewers. All outcomes reported in the included studies were systematically evaluated to determine suitability for meta-analysis. Random-effects models and sensitivity analyses were used to account for interstudy variation. Renal stone incidence rates were pooled to generate summary proportions. Risk of bias assessment was completed using the Risk of Bias Assessment tool for Non-randomized Studies. RESULTS: Twenty-one studies were included through systematic review and approximately 4.7 million pregnancies across nine studies were included for meta-analysis. There are three major findings of this review regarding renal stone incidence in pregnancy and maternal, child, and birth-related outcomes associated with renal stones. First, we found pooled incidence of renal stones was 0.49%, or one case for every 204 pregnancies. Second, renal stones during pregnancy were significantly associated with the development of preeclampsia and urinary tract infection, as well as increased likelihood of low birth weight, preterm labor, and C-section deliveries. However, renal stones were not significantly associated with premature rupture of membranes or infant mortality. Third, there were limited obstetrical complications reported with either medical or surgical therapies although comparative outcomes were not provided in the majority of studies, precluding formal meta-analysis. CONCLUSIONS: Although renal stones in pregnancy are relatively rare, there may be an associated risk of serious adverse obstetrical outcomes. However, further research is required to understand whether these obstetrical outcomes are causal or due to other confounders. Interdisciplinary care and pregnancy-specific counseling should be advised for pregnant women with kidney stones.

A Review of the Evidence and Current Applications of Portable Translingual Neurostimulation Technology.

OBJECTIVES: Translingual neurostimulation (TLNS) with adjunct physical rehabilitation is used to treat balance and gait deficits in several chronic neurological conditions. The purpose of this review is to summarize and appraise the evidence currently available on the portable TLNS device and to assess its potential clinical application. MATERIALS AND METHODS: In this narrative review, MEDLINE, EMBASE, Web of Science, and Google Scholar were searched for primary research investigating the use of portable TLNS devices on any neurologic condition. Data were extracted, reviewed, and appraised with respect to study design, conduct, and reporting. RESULTS: Five randomized controlled trials (RCTs), three quasi-experimental trials, and seven case reports/series were found. Most studies demonstrated improvements in balance and gait deficits secondary to traumatic brain injury and multiple sclerosis, but evidence is also present to a lesser degree for stroke and balance disorder patients. In these studies, the feasibility and safety of TLNS have been convincingly demonstrated. Functional magnetic resonance studies have also suggested a plausible neuroplastic therapeutic mechanism. However, the efficacy of TLNS remains unclear due to bias and confounding within studies, and heterogeneity of results between studies. CONCLUSIONS: TLNS is a promising treatment modality for various chronic neurological conditions that are often refractory to conventional therapy. However, TLNS technology remains largely investigational as high-quality RCTs are still required to elucidate efficacy, optimal dosages, necessary treatment durations, and treatment durability. Further research to develop an appropriate control group is needed for scientifically valid comparisons of TLNS.

The association between renal stones and fasting: A systematic review.

INTRODUCTION: Fasting is a common cultural practice worldwide for both religious and dietary reasons. However, there is concern that fasting may be a risk factor for the development of renal stones. To date, there has not been a systematic assessment of the literature regarding the association between renal stones and fasting. METHODS: We conducted a systematic review following PRISMA guidelines of three databases: Medline-OVID, EMBASE, and CINAHL. All screening and extraction was completed in parallel with two independent reviewers. RESULTS: Of the 1501 database citations, a total of 10 observational studies with a total of 9906 participants were included. Nine of the studies were conducted in the context of Islamic fasting during Ramadan, with the majority (7/9) finding that renal colic incidence was unaffected by the month of fasting. In contrast, two studies noted an increased incidence among fasting populations. Two other studies noted that urine metabolites and density were altered with fasting but did not translate into clinical outcomes. CONCLUSIONS: Based on the available evidence, it is unlikely that fasting significantly increases the risk of renal stones. Physicians should counsel higher-risk patients on safe fasting practices.

Predictors of participant nonadherence in lung cancer screening programs: a systematic review and meta-analysis.

The low nonadherence rates reported by large low-dose computed tomography (LDCT) lung cancer screening trials were not necessarily replicated outside of trial conditions. This systematic review and meta-analysis identified predictors of participant nonadherence to returning for annual LDCT screening. The systematic review protocol was registered at PROSPERO (CRD42019118347). MEDLINE, EMBASE, CINAHL, AgeLine, grey literature sources, and reference lists of included studies were searched until March 1st, 2020. Primary research articles were eligible for inclusion if they screened current or former smokers using LDCT as their primary screening modality and reported on participant demographics or programmatic interventions that predicted nonadherence. Risk of bias assessment was performed at both study and outcome levels. The primary outcome was predictors of nonadherence. The secondary outcomes were relative risks (RR) of second round nonadherence based on identified predictors, which were calculated using random-effects meta-analyses. Across 13 included studies (total n = 15,790; range: 157-3642), the overall rate of nonadherence was 28% (95% CI: 20-37%). Studies identified greater nonadherence in participants younger than 60 or older than 74, with longer travel distances to screening centers, and having a low risk perception of lung cancer. Meta-analyses identified higher nonadherence in community-based compared to academic-based programs, but this did not reach significance (32% versus 27%; p = 0.32). Current smokers were more likely to be nonadherent compared to former smokers (RR 1.23, 95% CI: 1.09-1.40; p < 0.01) while white participants were less likely nonadherent compared to non-white participants (RR 0.69, 95% CI: 0.60-0.81; p < 0.0001). No differences existed between male and female participants (RR 0.99, 95% CI: 0.85-1.15; p = 0.85). Programmatic interventions, including dedicated program coordinators, reminder calls/letters, and mobile LDCT scanners reduced nonadherence in lung cancer screening programs. These interventions should be targeted/tailored toward the subpopulations with the highest nonadherence rates.

Development of pulmonary endovascular metastases following vertebroplasty: case report.

A 69-year-old man developed pulmonary metastases following vertebroplasties for pathological fractures of vertebrae T12-L4. The fractures developed due to spinal metastases from castrate-resistant prostate cancer. A CT scan performed 1 month prior indicated no evidence of pulmonary malignancy. However, CT scans performed 2 months after the vertebroplasties demonstrated intravascular pulmonary metastases distributed similarly to embolized polymethylmethacrylate. Vertebroplasty is a well-established procedure for symptomatic management of vertebral compression fractures. However, studies have demonstrated an increase in circulating tumor cells following vertebroplasties, theoretically increasing the risk of distant metastases. In this case, the chronicity and radiological findings suggest that the pulmonary intravascular metastases may have resulted from the vertebroplasties.

Outcomes of Long-term Interval Rescreening With Low-Dose Computed Tomography for Lung Cancer in Different Risk Cohorts.

INTRODUCTION: We hypothesize that the incidence of screen-detected lung cancer (LC), in participants with previously negative scans, will be highest in the cohort with the highest baseline risk score. METHODS: Individuals with negative baseline screening results from the Princess Margaret International Early Lung Cancer Action Program before 2009 underwent low-dose computed tomography rescreening from 2015 to 2018. Individuals were contacted in order of descending risk, as determined by the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial's PLCOM2012 6-year LC risk-prediction model, and then categorized into three risk cohorts according to their baseline risks. The incidence of LC in each risk cohort was determined and compared. Chi-square testing was used for categorical variables and one-way analysis of variance on ranks was used for continuous variables. RESULTS: Of the 1261 participants we attempted to re-contact, 359 participants returned for a rescreening scan (mean of 7.6 years between scans). Participants were divided into low (<2%), moderate (≥2% to <3.5%), and high baseline risk (≥3.5%) cohorts. On average, those in the high-risk cohort compared to the moderate- and low-risk cohorts were older (66 years versus 62 and 59 years) and had a greater smoking history (54 pack-years versus 47 and 29 pack-years). The incidence of cancer in the high-risk cohort was significantly higher than in the moderate-risk cohort (11% versus 1.7%, p = 0.002). CONCLUSIONS: There was a significantly higher incidence of LC in the high-risk cohort than in the moderate-risk cohort. The cut-point between the high- and moderate-risk was determined to be greater than or equal to 3.5% of the 6-year baseline risk.