Background: To evaluate if United States Medical Licensing Examination (USMLE) Step 1, USMLE Step 2 CK, USMLE Step 3, and residency third-year in-service training exam (ITE) scores predict the results of American Board of Internal Medicine Certifying Exam (ABIM-CE). Methods: We performed a retrospective review of USMLE Step 1, USMLE Step 2 CK, USMLE Step 3, third-year residency ITE scores and ABIM-CE results of IM residents at our residency program from 2004 through 2017. Statistical analysis was perfrormed using Pearson correlation coefficient, and logistic regression to assess the relationship between USMLE Step 1, USMLE Step 2CK, USMLE Step 3, 3rd year ITE scores and ABIM-CE results. We used Multivariate logistic regression to predict pass or fail results in ABIM-CE based on USMLE and third-year ITE test scores controlling for other covariates. Results: Among 114 Internal Medicine MD residents included in the study, 92% (n = 105) passed the ABIM-CE. The OR of passing ABIM-CE was 2.70 (95% CI = 1.38–5.29), 2.31 (95% CI = 1.33–4.01), and 1.63 (95% CI = 0.81–3.29) with a ten-point increase in USMLE Step 1, USMLE Step 2 CK and USMLE Step 3 scores respectively. The OR of ABIM-CE passing chance was 2.96 (95% CI = 0.95–9.20), with a ten-point increase in the average score of the above three exams. A 5 % increase in ITE percentage raised the likelihood of passing ABIM-CE (OR 2.92, 95% CI 1.15–7.38). All residents who failed ABIM-CE had Step 1 scores < 220. Among 31 residents with Step 2 CK score < 220, 20% (n = 6) failed ABIM. Similarly, 9% of residents with USMLE Step 3 score < 220 failed ABIM-CE; all residents who failed had scored < 220. The probability curve predicted that the chance of passing ABIM- CE was around 80% with USMLE scores greater than 200 and increased to almost 100% with USMLE scores of 250 or more. Conclusion: USMLE Step 1, USMLE Step 2 CK, and third-year ITE scores can predict the chances of passing ABIM-CE. The third-year ITE score has a higher preditive value compared to USMLE Step 1 and USMLE Step 2 scores. USMLE Step 1 scores more predictive of ABIM-CE results compared to USMLE Step 2CK scores. Thus, residency programs can identify internal medicine residents at risk of failing ABIM-CE and formulate interventions at an early stage during residency training. Measures such as enrolling them in question banks or board review courses can be helpful in improving their chances of passing ABIM-CE.

Background: Medical students must meet curricular expectations and pass national licensing examinations to become physicians. However, no previous studies explicitly modeled stages of medical students acquiring basic science knowledge. In this study, we employed an innovative statistical model to characterize students’ growth using progress testing results over time and predict licensing examination performance. Methods: All students matriculated from 2016 to 2017 in our medical school with USMLE Step 1 test scores were included in this retrospective cohort study (N = 358). Markov chain method was employed to: 1) identify latent states of acquiring scientific knowledge based on progress tests and 2) estimate students’ transition probabilities between states. The primary outcome of this study, United States Medical Licensing Examination (USMLE) Step 1 performance, were predicted based on students’ estimated probabilities in each latent state identified by Markov chain model. Results: Four latent states were identified based on students’ progress test results: Novice, Advanced Beginner I, Advanced Beginner II and Competent States. At the end of the first year, students predicted to remain in the Novice state had lower mean Step 1 scores compared to those in the Competent state (209, SD = 14.8 versus 255, SD = 10.8 respectively) and had more first attempt failures (11.5% versus 0%). On regression analysis, it is found that at the end of the first year, if there was 10% higher chance staying in Novice State, Step 1 scores will be predicted 2.0 points lower (95% CI: 0.85–2.81 with P < .01); while 10% higher chance in Competent State, Step 1scores will be predicted 4.3 points higher (95% CI: 2.92–5.19 with P < .01). Similar findings were also found at the end of second year medical school. Conclusions: Using the Markov chain model to analyze longitudinal progress test performance offers a flexible and effective estimation method to identify students’ transitions across latent stages for acquiring scientific knowledge. The results can help identify students who are at-risk for licensing examination failure and may benefit from targeted academic support.

Mendelian Randomization (MR) is a widely embraced approach to assess causality in epidemiological studies. Two-stage least squares (2SLS) method is a predominant technique in MR analysis. However, it can lead to biased estimates when instrumental variables (IVs) are weak. Moreover, the issue of the winner’s curse could emerge when utilizing the same dataset for both IV selection and causal effect estimation, leading to biased estimates of causal effects and high false positives. Focusing on one-sample MR analysis, this paper introduces a novel method termed Mendelian Randomization with adaptive Sample-sPLitting with cross-fitting InstrumenTs (MR-SPLIT), designed to address bias issues due to IV selection and weak IVs, under the 2SLS IV regression framework. We show that the MR-SPLIT estimator is more efficient than its counterpart cross-fitting MR (CFMR) estimator. Additionally, we introduce a multiple sample-splitting technique to enhance the robustness of the method. We conduct extensive simulation studies to compare the performance of our method with its counterparts. The results underscored its superiority in bias reduction, effective type I error control, and increased power. We further demonstrate its utility through the application of a real-world dataset. Our study underscores the importance of addressing bias issues due to IV selection and weak IVs in one-sample MR analyses and provides a robust solution to the challenge.

The availability of local, state, and national data on alcohol outlet density have important implications for policies and interventions aiming to reduce alcohol-related problems. High-quality data...