Search

It has been well established that randomized clinical trials have poor external validity, resulting in findings that may not apply to relevant—or target—populations. When the trial is sampled from the target population, generalizability methods have been proposed to address the applicability of trial findings to target populations. When the trial sample and target populations are distinct, transportability methods may be applied for this purpose. However, generalizability and transportability studies present challenges, particularly around the strength of their conclusions. We review and summarize state-of-the-art methods for translating trial findings to target populations. We additionally provide a novel step-by-step guide to address these challenges, illustrating principles through a published case study. When conducted with rigor, generalizability and transportability studies can play an integral role in regulatory decisions by providing key real-world evidence. [ABSTRACT FROM AUTHOR]

According to recent census data, Hispanic and Latino populations comprise the largest minority group in the United States. Despite ongoing efforts for improved diversity, equity, and inclusion, Hispanics remain underrepresented in medicine (UIM). In addition to well-established benefits to patient care and health systems, physician diversity and increased representation in academic faculty positively impact the recruitment of trainees from UIM backgrounds. Disproportionate representation (as compared to increases of certain underrepresented groups in the US population) has direct implications for recruitment of UIM trainees to residency programs. To examine the number of full-time US medical school faculty physicians who self-identify as Hispanic in light of the increasing Hispanic population in the United States. We analyzed data from the Association of American Medical Colleges from 1990 to 2021, looking at those academic faculty who were classified as Hispanic, Latino, of Spanish Origin, or of Multiple Race—Hispanic. We used descriptive statistics and visualizations to illustrate the level of representation of Hispanic faculty by sex, rank, and clinical specialty over time. Overall, the proportion of faculty studied who identified as Hispanic increased from 3.1% (1990) to 6.01% (2021). Moreover, while the proportion of female Hispanic academic faculty increased, there remains a lag between females versus males. Our analysis shows that the number of full-time US medical school faculty who self-identify as Hispanic has not increased, though the population of Hispanics in the United States has increased. [ABSTRACT FROM AUTHOR]

Inverse odds of participation weighting (IOPW) has been proposed to transport clinical trial findings to target populations of interest when the distribution of treatment effect modifiers differs between trial and target populations. We set out to apply IOPW to transport results from an observational study to a target population of interest. We demonstrated the feasibility of this idea with a real-world example using a nationwide electronic health record derived de-identified database from Flatiron Health. First, we conducted an observational study that carefully adjusted for confounding to estimate the treatment effect of fulvestrant plus palbociclib relative to letrozole plus palbociclib as a second-line therapy among estrogen receptor (ER)-positive, human epidermal growth factor receptor (HER2)-negative metastatic breast cancer patients. Second, we transported these findings to the broader cohort of patients who were eligible for a first-line therapy. The interpretation of the findings and validity of such studies, however, rely on the extent that causal inference assumptions are met. [ABSTRACT FROM AUTHOR]

To address potential safety concerns of Janus Kinase Inhibitors (JAK-Is), we characterized their safety profile in the atopic dermatitis (AD) patient population. In this retrospective observational study, we used propensity score-based methods and a Poisson modeling framework to estimate the incidence of health outcomes of interest (HOI) for the AD patient. To that end, two mutually exclusive cohorts were created using a real world data resource: a rheumatoid arthritis (RA) cohort, where we directly quantify the safety risk of JAK-Is on HOIs, and an AD cohort, that comprises the target population of interest and to whom we transport the results obtained from the RA cohort. The RA cohort included all adults who filled at least one prescription for a JAK-I (tofacitinib, baricitinib, or upadacitinib) between 1 January 2017 and 31 January 2020. The AD cohort consisted of all adults diagnosed with AD during the same period. We first estimated the incidence rate of each HOI in the RA cohort, and then transported the results to the AD population. The RA and AD cohorts included 5,296 and 261,855 patients, respectively. On average, patients in the AD cohort were younger, more often male, more likely to be Asian, and had higher household income. They also had a lower prevalence of several comorbid conditions including hypertension, chronic kidney disease, obesity, and depression. Overall, the transported incidence rates of the HOIs to the AD cohort were lower than those obtained in the RA cohort by 13–50%. We applied transportability methods to characterize the risk of the HOIs in the AD population and found absolute risks higher than that of the general population. Future work is needed to validate these conclusions in comparable populations. [ABSTRACT FROM AUTHOR]

Background: The Society for Academic Emergency Medicine (SAEM) has a core value to promote a diverse workforce for patients, providers, and learners. Understanding the organization's membership demographics and how that compares to the academic emergency medicine (EM) workforce is prerequisite to the success of this core value. Methods: We obtained 2020 faculty membership data sets from the Association of American Medical Colleges (AAMC) and SAEM; data included self‐reported sex, race and ethnicity, and academic rank (professor, associate professor, assistant professor, and instructor). We employed standardized mean difference (SMD) to quantify difference in proportions between data sets. Results: We identified 5874 (AAMC) and 2785 (SAEM) faculty. The AAMC (38.3%) and the SAEM (41.3%) had similar proportions of overall female faculty (SMD 0.063) although SAEM (compared to AAMC) had a higher proportion of female full (25.5% vs. 20.5%, SMD 0.121) and assistant (46.5% vs. 41.2%, SMD 0.106) professors. With the exception of Hispanic instructors, SAEM (compared to AAMC) also had higher proportions of Black and Hispanic female faculty at all ranks (SMD ranging from 0.109 to 0.777). Conclusion: SAEM faculty demographics generally reflect that of the academic EM workforce demographics reported in the AAMC database and that overall, the proportions of female, Black, and Hispanic faculty in SAEM are slightly larger than those in the AAMC database. However, faculty who identify as Black or Hispanic in both the AAMC and the SAEM databases (compared to the overall U.S. population) are dramatically underrepresented. [ABSTRACT FROM AUTHOR]

Background: Understanding factors that impact tissue oxygen extraction may guide red blood cell (RBC) transfusion decision making in preterm infants. Our objective was to assess the influence of enteral feeding and anemia on cerebral and mesenteric oxygen saturation (Csat and Msat) and fractional tissue oxygen extraction (cFTOE and mFTOE) over the entire time course of RBC transfusion.Study Design and Methods: Preterm, very low-birth-weight infants receiving RBC transfusions at a single center were enrolled. Near-infrared spectroscopy sensors measured Csat and Msat levels from an hour before transfusion to 24 hours after. During this period, changes in Csat, Msat, cFTOE, and mFTOE were described, and their association with enteral feeding status and pretransfusion degree of anemia were assessed using generalized estimating equations.Results: RBC transfusion data from 31 preterm infants were included. Infants receiving enteral feeds exhibited lower pretransfusion Msat. Infants with pretransfusion hematocrit greater than 30% exhibited higher pretransfusion Csat and lower pretransfusion cFTOE. Such differences in baseline measurements persisted through 24 hours after transfusion. However, no statistically significant differences in oxygenation measures over time by enteral feeding or anemia status were identified.Conclusion: Compared to NPO, enteral feeding was associated with lower Msat; anemia (hematocrit ≤30%) was associated with lower Csat and higher cFTOE. Over the time course of RBC transfusion, trajectories of Csat, Msat, cFTOE and mFTOE did not differ by enteral feeding or anemia status. [ABSTRACT FROM AUTHOR]

Background: Hirschsprung's disease, or congenital aganglionosis, is a developmental disorder of the enteric nervous system and is the most common cause of intestinal obstruction in neonates and infants. The disease has more than 80% heritability, including significant associations with rare and common sequence variants in genes related to the enteric nervous system, as well as with monogenic and chromosomal syndromes.Methods: We genotyped and exome-sequenced samples from 190 patients with Hirschsprung's disease to quantify the genetic burden in patients with this condition. DNA sequence variants, large copy-number variants, and karyotype variants in probands were considered to be pathogenic when they were significantly associated with Hirschsprung's disease or another neurodevelopmental disorder. Novel genes were confirmed by functional studies in the mouse and human embryonic gut and in zebrafish embryos.Results: The presence of five or more variants in four noncoding elements defined a widespread risk of Hirschsprung's disease (48.4% of patients and 17.1% of controls; odds ratio, 4.54; 95% confidence interval [CI], 3.19 to 6.46). Rare coding variants in 24 genes that play roles in enteric neural-crest cell fate, 7 of which were novel, were also common (34.7% of patients and 5.0% of controls) and conferred a much greater risk than noncoding variants (odds ratio, 10.02; 95% CI, 6.45 to 15.58). Large copy-number variants, which were present in fewer patients (11.4%, as compared with 0.2% of controls), conferred the highest risk (odds ratio, 63.07; 95% CI, 36.75 to 108.25). At least one identifiable genetic risk factor was found in 72.1% of the patients, and at least 48.4% of patients had a structural or regulatory deficiency in the gene encoding receptor tyrosine kinase (RET). For individual patients, the estimated risk of Hirschsprung's disease ranged from 5.33 cases per 100,000 live births (approximately 1 per 18,800) to 8.38 per 1000 live births (approximately 1 per 120).Conclusions: Among the patients in our study, Hirschsprung's disease arose from common noncoding variants, rare coding variants, and copy-number variants affecting genes involved in enteric neural-crest cell fate that exacerbate the widespread genetic susceptibility associated with RET. For individual patients, the genotype-specific odds ratios varied by a factor of approximately 67, which provides a basis for risk stratification and genetic counseling. (Funded by the National Institutes of Health.). [ABSTRACT FROM AUTHOR]

This study examines differences in proportions of US medical school faculty who self-identified as Black or African American between 1990 and 2020 based on data from the Association of American Medical Colleges Faculty Roster. [ABSTRACT FROM AUTHOR]

Background: An early HIV diagnosis improves patient outcomes, reduces the burden of undiagnosed HIV, and limits transmission. There is a need for an updated assessment of HIV testing rates in the emergency department (ED)., Setting: The National Hospital Ambulatory Medical Care Survey sampling ED visits were weighted to give an estimate of ED visits across all US states in 2018., Methods: We analyzed patients aged 13-64 years without known HIV and estimated ED visits with HIV testing and then stratified by race, ethnicity, and region. Descriptive statistics and mapping were used to illustrate and compare patient, visit, and hospital characteristics for visits with HIV testing., Results: Of 83.0 million weighted visits to EDs in 2018 by patients aged 13-64 years without a known HIV infection (based on 13,237 National Hospital Ambulatory Medical Care Survey sample visits), HIV testing was performed in 1.05% of visits. HIV testing was more frequent for patients aged 13-34 years compared with that for patients aged 35-64 years (1.32% vs. 0.82%, P = 0.056), Black patients compared with that for White and other patients (1.73% vs. 0.79% and 0.41%, P = 0.002), Hispanic or Latino patients compared with that for non-Hispanic or Latino patients (2.18% vs. 0.84%, P = 0.001), and patients insured by Medicaid compared with that for patients insured by private or other insurance (1.71% vs. 0.64% and 0.96%, P = 0.003). HIV testing rates were the highest in the Northeast (1.72%), followed by the South (1.05%)., Conclusions: HIV testing occurred in a minority of ED visits. There are differences in rates of HIV testing by race, ethnicity, and location. Although rates of testing have increased, rates of ED-based HIV testing remain low., Competing Interests: The authors have no conflicts of interest to disclose., (Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.)

Objectives: Most population-based cancer databases lack information on metastatic recurrence. Electronic medical records (EMR) and cancer registries contain complementary information on cancer diagnosis, treatment and outcome, yet are rarely used synergistically. To construct a cohort of metastatic breast cancer (MBC) patients, we applied natural language processing techniques within a semisupervised machine learning framework to linked EMR-California Cancer Registry (CCR) data., Materials and Methods: We studied all female patients treated at Stanford Health Care with an incident breast cancer diagnosis from 2000 to 2014. Our database consisted of structured fields and unstructured free-text clinical notes from EMR, linked to CCR, a component of the Surveillance, Epidemiology and End Results Program (SEER). We identified de novo MBC patients from CCR and extracted information on distant recurrences from patient notes in EMR. Furthermore, we trained a regularized logistic regression model for recurrent MBC classification and evaluated its performance on a gold standard set of 146 patients., Results: There were 11 459 breast cancer patients in total and the median follow-up time was 96.3 months. We identified 1886 MBC patients, 512 (27.1%) of whom were de novo MBC patients and 1374 (72.9%) were recurrent MBC patients. Our final MBC classifier achieved an area under the receiver operating characteristic curve (AUC) of 0.917, with sensitivity 0.861, specificity 0.878, and accuracy 0.870., Discussion and Conclusion: To enable population-based research on MBC, we developed a framework for retrospective case detection combining EMR and CCR data. Our classifier achieved good AUC, sensitivity, and specificity without expert-labeled examples., (© The Author(s) 2019. Published by Oxford University Press on behalf of the American Medical Informatics Association.)

Over 75 million Americans have multiple concurrent chronic conditions and medical decision making for these patients is mostly based on retrospective cohort studies. Current methods to generate cohorts of patients with comorbidities are neither scalable nor generalizable. We propose a supervised machine learning algorithm for learning comorbidity phenotypes without requiring manually created training sets. First, we generated myocardial infarction (MI) and type-2 diabetes (T2DM) patient cohorts using ICD9-based imperfectly labeled samples upon which LASSO logistic regression models were trained. Second, we assessed the effects of training sample size, inclusion of physician input, and inclusion of clinical text features on model performance. Using ICD9 codes as our labeling heuristic, we achieved comparable performance to models created using keywords as labeling heuristic. We found that expert input and higher training sample sizes could compensate for the lack of clinical text derived features. However, our best performing model included clinical text as features with a large training sample size.