Yasser Albogami

Background: Data linkage in pharmacoepidemiological research is commonly employed to ascertain exposures and outcomes or to obtain additional information on confounding variables. However, to protect patient confidentiality, unique patient identifiers are not provided, which makes data linkage across multiple sources challenging. The Saudi Real-World Evidence Network (SRWEN) aggregates electronic health records from various hospitals, which may require robust linkage techniques. Objective: We aimed to evaluate and compare the performance of deterministic, probabilistic, and machine learning (ML) approaches for linking deidentified data of patients with multiple sclerosis (MS) from the SRWEN and Ministry of National Guard Health Affairs electronic health record systems. Methods: A simulation-based validation framework was applied before linking real-world data sources. Deterministic linkage was based on predefined rules, whereas probabilistic linkage was based on a similarity score–based matching. For ML, both similarity score–based and classification approaches were applied using neural networks, logistic regression, and random forest models. The performance of each approach was assessed using confusion matrices, focusing on sensitivity, positive predictive value, F ₁ score, and computational efficiency. Results: The study included linked data of 2247 patients with MS from 2016 to 2023. The deterministic approach resulted in an average F ₁ score of 97.2% in the simulation and demonstrated varying match rates in real-world linkage: 1046/2247 (46.6%) to 1946/2247 (86.6%). This linkage was computationally efficient, with run times of F ₁ score of 93.9% in the simulation, with real-world match rates ranging from 1472/2247 (65.5%) to 2144/2247 (95.4%) and processing times ranging from approximately 0.1 to 5 seconds per rule. ML approaches achieved high performance ( F ₁ score reached 99.8%) but were computationally expensive. Processing times ranged from approximately 13 to 16,936 seconds for the classification-based approaches and from approximately 13 to 7467 seconds for the similarity score–based approaches. Real-world match rates from ML models were highly variable depending on the method used; the similarity score–based approach identified 789/2247 (35.1%) matched pairs, whereas the classification-based approach identified 2014/2247 (89.6%). Conclusions: Probabilistic linkage offers high linkage capacity by recovering matches missed by deterministic methods and proved to be both flexible and efficient, particularly in real-world scenarios where unique identifiers are lacking. This method achieved a great balance between recall and precision, enabling better integration of various data sources that could be useful in MS research.

Aim: Preclinical studies have suggested that glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) may induce thyroid gland hyperplasia, raising concerns about potential thyroid‐related risks. Given their increasing use and the limited evidence on thyroid safety, this study assessed the association between GLP‐1RA use and the incidence of hypothyroidism using real‐world data. Methods: We conducted an active‐comparator, new‐user cohort study using the Real‐World Evidence Research Network (SRWEN) (2016–2023). Adults (≥ 18 years) initiating GLP‐1RAs or dipeptidyl peptidase‐4 inhibitors (DPP‐4is) were followed from first prescription until hypothyroidism, treatment discontinuation, switching, death, or study end. Hypothyroidism was identified through ICD‐10 codes or levothyroxine prescriptions. Inverse probability of treatment weighting was applied to adjust for confounding, and weighted Cox proportional hazards models were used to estimate hazard ratios (HRs). RStudio 4.4.0 was used for analyses. Results: A total of 47 017 patients were included (6800 GLP‐1RA users; 40 217 DPP‐4i users). GLP‐1RA users were younger (mean age 50 vs. 58 years) and more often female. The incidence rate of hypothyroidism was 128 per 10 000 person‐years in GLP‐1RA users compared to 150 per 10 000 person‐years in DPP‐4i users. GLP‐1RA use was not associated with a statistically significant risk of hypothyroidism (adjusted HR 1.04, 95% CI 0.69–1.57). Sensitivity analyses extending follow‐up by 30 and 60 days yielded consistent findings. Conclusion: In this real‐world analysis, GLP‐1RA use was not associated with an increased incidence of hypothyroidism compared to DPP‐4is. Findings were consistent across sensitivity and subgroup analyses. Although findings do not suggest a short‐term risk, longer‐term studies are warranted to further evaluate thyroid safety.

Introduction: The aim of this study is to use observational methods to evaluate reliability of evidence generated by a study of the effect of glucagon‐like peptide 1 receptor agonists (GLP‐1RA) on chronic lower respiratory disease (CLRD) outcomes among Type‐2 diabetes mellitus (T2DM) patients.Research Design and Methods: We independently reproduced a study comparing effects of GLP‐1RA versus dipeptidyl peptidase‐4 inhibitors (DPP4‐i) on CLRD outcomes among patients with T2DM and prior CLRD. We reproduced inputs and outputs using the original study data (national administrative claims) and evaluated the robustness of results in comparison to alternate design/analysis decisions. To evaluate generalizability, we applied an analysis protocol and conducted a meta‐analysis across a research network that includes a diverse array of populations and data sources. We also produced additional analyses evaluating individual drugs within the GLP‐1RA class and CLRD outcomes.Results: We confirmed alignment of study inputs and outputs and closely reproduced effect estimates and sensitivity analyses. Adjusted effect estimates were robust to empirical calibration. Network meta‐analysis confirmed original findings but indicated weaker effects than originally published. Meta‐analysis of drugs within the GLP‐1RA class against DPP4‐i provided evidence that effects vary within the GLP‐1RA class, indicating stronger effects for exenatide and weaker effects of dulaglutide.Conclusions: This study supports and establishes the reliability of the original study by (1) producing consistent findings in a range of alternate databases and populations, (2) demonstrating effects for multiple drugs within the GLP‐1RA class, and (3) independently confirming the reproducibility of the original study and its findings. This reliability evaluation provides the beginnings of a broader framework for using standardized tools and distributed data networks to systematically interrogate the reliability of findings generated using observational data.

Introduction: Use observational methods to evaluate reliability of evidence generated by a study of the effect of glucagon-like peptide 1 receptor agonists (GLP-1RA) on chronic lower respiratory disease (CLRD) outcomes among type-2 diabetes mellites (T2DM) patients. Research Design and Methods: We independently reproduced a study comparing effects of GLP-1RA versus dipeptidyl peptidase-4 inhibitors (DPP4-i) on CLRD outcome among patients with T2DM and prior CLRD. We reproduced inputs and outputs using the original study data (national administrative claims) and evaluated robustness to alternate design/analysis decisions. To evaluate generalizability, we applied the protocol and meta-analyzed across a research network including diverse array of populations and data sources. We also produced additional analyses evaluating individual drugs within the GLP-1RA class. Results: We confirmed alignment of study inputs and outputs and closely reproduced effect estimates and sensitivity analyses. Adjusted effect estimates were robust to empirical calibration. Network meta-analysis confirmed original findings, but indicated weaker effects than originally published. Meta-analyzing drugs within the GLP-1RA class against DPP4-I provided some evidence that effects vary within the GLP-1RA class, indicating stronger effects for exenatide and weaker effects of dulaglutide. Conclusions: This study supports the reliability of the original study by 1) confirming the findings in a range of alternate databases and populations 2) demonstrating effects for multiple drugs within the GLP-1RA class, and 3) independently confirming the reproducibility original study and its findings. We propose that clinicians treating patients with T2DM and a history of CLRD consider GLP-1RA in absence of strong motivating reasons to select another therapy.

Background: Despite its high efficacy in treating severe acne, isotretinoin is associated with serious side effects, including teratogenicity. However, the extent of isotretinoin exposure during pregnancy in Saudi Arabia remains unknown. Objectives: This study aims to quantify the extent of fetal exposure to isotretinoin in Saudi Arabia and to evaluate adherence to risk minimization measures approved by the Saudi Food and Drug Authority. Design: Retrospective cohort study. Methods: This multicenter retrospective study included a cohort of 6233 women of childbearing ages (WCBAs) who had received isotretinoin therapy between 2015 and 2020. Exposure to isotretinoin use was ascertained from patients’ electronic health records and was defined as any positive pregnancy test (urine or serum) or any diagnosis or procedure related to pregnancy occurring during the risk period. We defined the risk period starting from isotretinoin initiation until up to 30 days after the last prescription. We quantified the overall incidence proportion of fetal exposure to isotretinoin by dividing the number of pregnancy cases during the risk period by the total study sample of WCBAs. Results: The cohort predominantly included young females (20–29 years), with a mean age of 24 years. Only 5% of the WCBAs used contraceptives, and 10% have a record of pregnancy testing. During the risk period, 34 pregnancies were identified, yielding a cumulative pregnancy incidence of 5.6 per 1000 WCBAs. Pregnancy outcomes for exposed women were about 5% of births had defects, while abortions accounted for 14.3% of pregnancies. Conclusion: Our investigation shows an alarming incidence of fetal exposure to isotretinoin in Saudi Arabia, substantially surpassing global estimates. These results underscore a critical need for enhanced interventions and robust risk minimization strategies tailored to the distinct challenges faced by the Saudi Arabian population.

Introduction: Pharmacists play a pivotal role in ensuring patients are administered safe and effective medications; however, they encounter obstacles such as elevated workloads and a scarcity of qualified professionals. Despite the prospective utility of large language models (LLMs), such as Generative Pre-trained Transformers (GPTs), in addressing pharmaceutical inquiries, their applicability in real-world cases remains unexplored. Objective: To evaluate GPT-based chatbots’ accuracy in real-world drug-related inquiries, comparing their performance to licensed pharmacists. Methods: In this cross-sectional study, authors analyzed real-world drug inquiries from a Drug Information Inquiry Database. Two independent pharmacists evaluated the performance of GPT-based chatbots (GPT-3, GPT-3.5, GPT-4) against human pharmacists using accuracy, detail, and risk of harm criteria. Descriptive statistics described inquiry characteristics. Absolute proportion comparative analyses assessed accuracy, detail, and risk of harm. Stratified analyses were performed for different inquiry types. Results: Seventy inquiries were included. Most inquiries were received from physicians (41%) and pharmacists (44%). Inquiries type included dosage/administration (34.2%), drug interaction (12.8%) and pregnancy/lactation (15.7%). Majority of inquires included adults (83%) and female patients (54.3%). GPT-4 had 64.3% completely accurate responses, comparable to human pharmacists. GPT-4 and human pharmacists provided sufficiently detailed responses, with GPT-4 offering additional relevant details. Both GPT-4 and human pharmacists delivered 95% safe responses; however, GPT-4 provided proactive risk mitigation information in 70% of the instances, whereas similar information was included in 25.7% of human pharmacists’ responses. Conclusion: Our study showcased GPT-4's potential in addressing drug-related inquiries accurately and safely, comparable to human pharmacists. Current GPT-4-based chatbots could support healthcare professionals and foster global health improvements.