Curso superior R1

Introduction to Evidence-Based Medicine

Overview of the Module

• Hernán Roben introduces himself as the educational coordinator and outlines his role in guiding a module on evidence-based medicine (EBM).

• He mentions that this module is part of a broader course on research introduction, which he will also teach.

Historical Context and Evolution

• Roben notes that EBM has been part of residency programs for many years but emphasizes that current practices differ significantly from those 20 years ago.

• He stresses the importance for healthcare professionals to understand and interpret scientific literature to avoid being scientifically illiterate.

Importance of Critical Reading

Why Engage in Critical Reading?

• Critical reading involves evaluating scientific literature rather than accepting it at face value; it's essential for assessing the validity of research findings.

• Roben highlights that many published papers aim to persuade rather than inform, particularly those funded by industry, necessitating careful evaluation by physicians.

Real-world Application

• He provides an example regarding bronchiolitis treatment with corticosteroids, illustrating how statistical significance does not always equate to clinical relevance or utility.

• The speaker points out a significant issue: 90% of research presented at conferences never gets published, limiting public awareness and access to important medical findings.

Understanding the Pressure to Publish in Medicine

The Motivation Behind Medical Publications

• The necessity of publishing papers is driven by a system that evaluates medical professionals based on their publication records, particularly in first-world countries like the United States.

• In Argentina, while the pressure may not be as intense, there is still an expectation for doctors to meet research goals during their contracts.

• Aspiring medical professionals are encouraged to publish early in their careers, as publications significantly enhance their prospects for residency and job positions.

The Quality of Published Research

• It is estimated that around 80% of published research does not contribute meaningful information; rather, it serves the need for doctors to maintain prestige and secure funding.

• Papers often highlight successes while obscuring failures; negative outcomes or patient losses are rarely reported transparently in publications.

Critical Reading of Research Papers

• Learning critical reading skills is essential for identifying key points and understanding what may be omitted from studies.

• Authors typically include all significant findings in their papers; if something isn't mentioned, it likely wasn't done correctly or at all.

Statistical Understanding in Medical Research

• Traditional education focuses heavily on p-values without adequately addressing their limitations regarding clinical impact.

• P-values indicate whether results could be due to chance but do not provide insight into how many patients will benefit from a treatment.

Practical Application of Statistical Measures

• The Number Needed to Treat (NNT) is introduced as a more practical measure than p-values; it indicates how many patients must be treated for one to experience a benefit.

• Understanding NNT helps clarify the real-world effectiveness of treatments beyond statistical significance.

External Validity and Real-world Implications

• External validity examines whether study results can be generalized outside controlled environments; real-life applications often yield less favorable outcomes than clinical trials.

• Patients involved in studies tend to adhere better to protocols compared to those receiving standard care outside trial conditions.

Case Study: Clinical Trials and Market Impact

• A case study involving Cell Therapeutics Inc. illustrates how clinical trial results can influence stock prices and market perceptions before drug approval.

The Impact of Clinical Research on Stock Prices

Overview of Pixantrone and Market Reactions

• The drug pixantrone's announcement led to a significant increase in stock prices, rising from $0.44 to over $0.80, effectively doubling the investment for those who acted quickly.

• A comparison is made with the Argentine market, noting that the financial movements surrounding this drug were larger than the entire Merbal index due to its smaller size.

FDA Committee Decision and Its Consequences

• Despite initial excitement, the FDA committee unanimously voted against approving pixantrone due to biases in research, causing stock prices to plummet again.

• This situation highlights the financial stakes behind clinical studies and how they can be influenced by industry interests and potential biases in research.

Understanding Evidence-Based Medicine (EBM)

Definition and Importance of EBM

• Evidence-based medicine (EBM) involves using the best scientific strategies available for making clinical decisions, emphasizing informed practice.

• Practitioners are encouraged to combine their limited experience with published evidence and patient preferences when making treatment decisions.

Patient Involvement in Treatment Decisions

• Historically, patient preferences were often overlooked; however, modern EBM emphasizes shared decision-making between doctors and patients.

• An example is given where a parent seeks help for a crying infant; practitioners must balance evidence with patient concerns about treatment options.

Navigating Treatment Options Based on Evidence

Evaluating Medications for Common Conditions

• There is some evidence suggesting certain medications may alleviate infant colic symptoms; however, no definitive conclusions exist regarding their effectiveness.

• Practitioners should communicate honestly with patients about medication efficacy while allowing them to make informed choices based on available evidence.

Vaccination Decisions: A Case Study

• Discussion around dengue vaccination reveals complexities in recommendations based on age groups and severity of disease risk.

• Patients are empowered to make choices regarding vaccinations after being informed about costs and benefits associated with different age demographics.

Understanding the Importance of Reading in Medical Practice

The Role of Intuition and Common Sense

• The speaker emphasizes the effectiveness of reading extensively, suggesting that intuition and common sense are innate qualities that cannot be taught in medical school.

• Not all doctors possess the same level of diagnostic intuition; some have a natural ability to understand complex cases better than others.

Continuous Learning and Evidence-Based Medicine

• Ongoing review of knowledge is crucial; one must not only learn but also revisit information regularly to stay updated.

• There is a significant delay (approximately 10 years) from research findings to their publication in textbooks, highlighting the need for direct engagement with primary sources like research papers.

Navigating Scientific Literature

• With over 2 million scientific publications released annually, distinguishing valuable studies from irrelevant ones is essential for effective practice.

• General practitioners should aim to read around 19 papers daily to keep up with relevant medical advancements, which is often impractical.

Reading Habits Among Medical Students

• A survey indicates that medical students typically dedicate only 1 to 2 hours per week to reading, with many not engaging at all during certain weeks.

• Post-residency physicians often struggle to find time for reading, with a significant percentage reporting no recent reading activity.

Embracing Questions in Clinical Practice

• Evidence-based medicine begins by questioning existing practices; many healthcare professionals fail to ask critical questions about treatment efficacy.

• The speaker encourages residents to maintain curiosity and continuously question their clinical decisions throughout their training.

Evaluating Research Papers Critically

• When reviewing research papers, three fundamental questions should always be asked regarding internal validity:

• Is the study well-designed?

• Was group assignment randomized?

• If randomization was not employed, it may invalidate the results.

Understanding Internal Validity in Research

Importance of Internal Validity

• The speaker emphasizes the significance of internal validity in research, highlighting that a well-conducted study should be randomized, controlled, and balanced without patient loss.

• After assessing the methodology, researchers must consider the practical implications for their patients, questioning if findings are applicable to local contexts like Argentina.

Critical Reading Resources

• A critical reading guide is available on the hospital's library page or course campus, summarizing how to read different types of papers effectively.

• Brian Haynes' quote underscores the transient nature of medical knowledge: "Half of what we teach will be incorrect in five years," stressing the need for continuous learning.

Evolving Medical Practices and Evidence

Historical Context

• The speaker recalls past practices such as administering ranitidine without sufficient evidence, illustrating how accepted treatments can change over time due to new findings.

• There was a long-standing belief in using ranitidine despite its eventual prohibition due to harmful effects; this serves as a caution against blindly following established norms.

Clinical Guidelines and Their Lifespan

• U.S. clinical practice guidelines introduced in 2004 have a validity period of five years unless confirmed or retracted sooner.

• This system ensures that outdated guidelines are regularly reviewed and updated based on new evidence.

The Role of Artificial Intelligence in Medicine

Introduction to AI

• The speaker introduces artificial intelligence (AI), describing it as an attempt to replicate human thought processes and problem-solving abilities.

Advancements in AI Applications

• Modern AI systems demonstrate improved learning capabilities over time, akin to software updates enhancing user experience with tools like Microsoft Word.

Practical Use Cases for AI

• As healthcare professionals begin utilizing language models trained through AI, they will enhance their daily practice by improving efficiency and accuracy over time.

Machine Learning vs. Deep Learning

Distinction Between Learning Types

• The speaker outlines two major fields within AI: machine learning (ML) and deep learning (DL), explaining their differences through medical examples.

Example Application: Machine Learning

• In ML applications for detecting lung cancer via CT scans, programmers set specific rules for identifying malignancies based on image characteristics.

Understanding Machine Learning and Deep Learning

Introduction to Reinforcement Learning

• The speaker introduces reinforcement learning, explaining that it improves over time through feedback, eventually outperforming human capabilities.

Deep Learning Explained

• Deep learning utilizes artificial neural networks designed to mimic the human brain's structure, processing information in a binary manner.

• The system identifies pathological patterns from numerous CT scans without predefined algorithms, leading to questions about its accuracy compared to human observation.

Challenges of Deep Learning in Medicine

• The concept of the "black box" is introduced, highlighting the difficulty in tracing how deep learning systems reach conclusions.

• Despite its advanced capabilities, deep learning is not widely used in medicine due to challenges in understanding decision-making processes.

Practical Applications of Machine Learning

• Everyday applications like Spotify utilize machine learning for personalized recommendations based on user behavior and preferences.

• Similar algorithms are employed by Netflix for movie recommendations, enhancing user experience through feedback mechanisms.

Transitioning to Advanced AI Systems

• The discussion shifts towards more complex AI systems such as Google Assistant and autonomous vehicles like Tesla.

• Emphasis is placed on the importance of providing clear prompts or instructions (referred to as "prompts") when interacting with AI models.

Crafting Effective Prompts for AI

• A prompt can vary significantly in length and complexity; effective communication with AI requires context and structured input.

• The process involves step-by-step interaction rather than overwhelming the system with information all at once.

Limitations and Real-world Testing of AI Capabilities

• The speaker shares an anecdote about testing an AI's ability to write a personal letter, illustrating both its potential and limitations.

• An example is provided where the AI generated a heartfelt letter using specific details requested by the speaker.

Understanding AI Hallucinations in Medicine

The Importance of Context in Communication

• The speaker reflects on personal happiness and acknowledges past shortcomings as a partner, promising to change and be more present.

• A metaphor is introduced about getting lost at the beach with a spouse, emphasizing how context can influence perception and understanding.

AI Hallucinations Explained

• The speaker draws parallels between personal anecdotes and medical scenarios, explaining that misinterpretations can lead to false beliefs or "hallucinations" in AI systems.

• It is highlighted that AI does not perceive the world like humans; rather, it operates through statistical processing without true intelligence.

Risks of Misinterpretation in Medical Practice

• Emphasizes the critical nature of accuracy in medicine, stating that even slight errors can have significant consequences.

• Acknowledges that while mistakes happen, there is an imperative to recognize when machines make errors or "hallucinate."

Real-world Examples of AI Errors

• Discusses a case where an AI system misread text and invented a non-existent scientific term ("microscopía electrónica vegetativa"), which was subsequently published in multiple papers.

• This incident illustrates how misinformation generated by AI can proliferate within academic literature.

Practical Applications of AI Tools

• Encourages participants to explore various applications for their devices that will aid them beyond the course context.

• Introduces ChatGPT as a powerful tool for reading academic papers but clarifies its limitations regarding patient-related inquiries.

Understanding GPT Technology

• Explains what GPT (Generative Pre-trained Transformer) technology is and its integration into various applications via APIs.

• Describes how users can find different GPT models within ChatGPT for specific tasks such as literature analysis or paper interpretation.

Utilizing GPT Effectively

• Highlights the importance of knowing what questions to ask GPT models since they are pre-programmed with knowledge relevant to specific fields.

• Mentions examples like Scholar GPT for literature analysis and Paper Interpreter for decoding academic works efficiently.

This structured overview captures key insights from the transcript while providing timestamps for easy reference.

Medical Coding and AI Tools

Understanding Medical Coding with AI

• The speaker introduces a medical coding tool called "medical coding GPT," which is programmed with the CE10 system to automatically code diagnoses like asthma without needing detailed prompts.

• The speaker emphasizes that this AI can perform tasks autonomously, acting as a virtual assistant for various medical inquiries and analyses.

• An example is provided where the speaker uses data from a 2021 Lancet paper on the Sputnik vaccine, demonstrating how the AI analyzes results and calculates risks associated with vaccination groups.

• The importance of understanding key metrics in research papers, such as risk attributable to vaccination and confidence intervals, is highlighted; these calculations are now performed by AI instead of manual methods.

• The speaker notes past inaccuracies in AI calculations but acknowledges improvements over time, indicating that current models provide more accurate results.

Comparison of Language Models

Introduction to Different Models

• Another language model named "Cloud" is introduced, which operates similarly to CHGPT but has stricter training regarding values and safety protocols.

• Grock is mentioned as another model capable of providing harmful information, contrasting it with Cloud's ethical guidelines against promoting negative behavior.

Practical Applications and Ethical Considerations

• A practical experiment was conducted where the speaker requested humorous insults (roasts) about a fictional character from different models. CHGPT provided a light-hearted roast while Cloud refused due to ethical concerns about bullying.

• This comparison illustrates how different models respond to similar prompts differently based on their programming ethics; Cloud promotes positive interactions rather than negative ones.

Critical Analysis Using Evidence-Based Guidelines

Application of Reading Guides

• The speaker discusses using Cloud for critical analysis of scientific articles based on reading guides developed from Sackett’s principles in evidence-based medicine.

• These guides help users systematically evaluate research quality and applicability in clinical practice.

Analysis of a Clinical Study on Dengue Vaccine

Introduction to the Historical Context

• The discussion begins with a reference to a historical collection of papers from the 1990s, emphasizing the importance of critical reading in understanding clinical studies.

• The speaker encourages Cloud to study specific guidelines and analyze a paper critically, demonstrating an interactive learning approach.

Critical Analysis of the Paper

• The focus shifts to analyzing a phase three study on a tetravalent dengue vaccine for children, highlighting its relevance and recent publication.

• Key aspects such as random assignment are discussed; it was stratified by region and age category, indicating proper randomization practices.

Importance of Patient Follow-Up

• Emphasizes that patient retention is crucial; losing more than 20% invalidates study results. A follow-up rate above 90% is considered excellent.

• The analysis notes that 97% of participants completed the study, showcasing effective patient management.

Understanding Intention-to-Treat Analysis

• Introduces the concept of "intention to treat" (ITT), explaining its significance in maintaining group integrity during analysis despite participant dropouts.

• ITT is contrasted with valid case analysis, which only considers those who received treatment. ITT reflects real-world scenarios better.

Practical Application and Ethical Considerations

• Discusses how researchers must consider adherence issues when evaluating treatment effectiveness based on protocol versus intention-to-treat analyses.

• A personal anecdote illustrates ethical dilemmas faced when recommending expensive treatments like bosoritide for patients in need.

Conclusion: Navigating Clinical Guidelines

• The speaker reflects on their extensive research into various clinical guidelines before making informed decisions about patient care, underscoring the responsibility healthcare professionals bear.

AI in Medical Research: A Case Study

The Efficacy of Bosoritide

• The speaker discusses a personal experience where AI quickly retrieved information about the efficacy of bosoritide for treating children with a specific condition, demonstrating the speed and efficiency of AI in research.

Technology Assessment in Healthcare

• Emphasizes the importance of analyzing available information based on clinical practice guidelines and technology assessment reports rather than academic papers to understand global recommendations regarding treatments.

Regulatory Insights

• Mentions findings from major regulatory bodies like EMA (European Medicines Agency) and NICE (National Institute for Health and Care Excellence), highlighting their significance in evaluating treatment options.

Cost-Effectiveness Analysis

• Discusses the necessity of conducting cost-effectiveness analyses, which can be complex but are crucial for understanding the economic implications of medical interventions.

Utilizing AI Tools

• Introduces various AI tools that assist in interpreting medical literature, including Sace and ChatGPT, showcasing how these technologies can simplify complex data into understandable formats.

Understanding Research Design

Types of Research Studies

• Outlines two main types of studies based on research objectives: descriptive studies that simply describe phenomena versus analytical or comparative studies that involve statistical analysis.

Experimental vs. Observational Studies

• Differentiates between experimental studies, where researchers intervene directly with subjects, and observational studies, where researchers merely observe outcomes without intervention.

Data Collection Timing

• Explains the distinction between retrospective studies (using previously collected data) and prospective studies (collecting new data moving forward), emphasizing methodological considerations related to bias and accuracy.

Methodological Challenges

• Highlights potential errors associated with retrospective data collection due to variability in how previous measurements were taken compared to standardized methods used in prospective studies.

Understanding Study Designs in Research

Importance of Study Design

• The speaker emphasizes the significance of study design, particularly the difference between retrospective and prospective studies. Prospective studies allow for greater precision in measurements.

Types of Studies: Longitudinal vs. Cross-sectional

• A distinction is made between longitudinal and cross-sectional studies. Longitudinal studies involve multiple measurements over time, while cross-sectional studies measure a variable at one point in time.

• An example is provided where measuring the weight of children upon admission to a hospital over a year would be considered a cross-sectional study since it captures data at one specific moment.

Measuring Variables

• In contrast, if glucose tolerance tests are conducted on children multiple times, this constitutes a longitudinal study due to repeated measurements of the same variable.

Experimental Design for Treatment Comparison

• The discussion shifts to an investigator's goal of comparing two medications for treating rheumatoid arthritis (RA). Participants are asked to consider what type of study design would be optimal for this comparison.

• The audience is engaged through voting on different options for study designs, highlighting the importance of understanding experimental versus observational designs.

Clinical Trials and Their Characteristics

• The speaker clarifies that comparing medication efficacy requires an experimental design, specifically mentioning clinical trials as a suitable approach.

• The term "clinical trial" is defined as a prospective, analytical, comparative study that can include various types of variables but fundamentally aims to assess treatment effectiveness.

Observational Studies: Smoking During Pregnancy

• A new scenario is introduced regarding research on smoking during pregnancy and its association with low birth weight. This sets up another opportunity for participants to identify the correct type of study design based on given options.

• Participants are prompted to analyze whether the study follows a comparative or observational framework by tracking outcomes among smokers versus non-smokers during pregnancy.

Clarifying Misconceptions About Study Types

• There’s confusion among participants about why certain options were chosen; some mistakenly identified longitudinal aspects when they should have recognized it as cross-sectional due to single measurement points like birth weight.

• The speaker reinforces that following subjects over time indicates longitudinal research but emphasizes that measuring outcomes like birth weight only once aligns with cross-sectional methodology.

This structured overview provides clarity on key concepts related to research methodologies discussed in the transcript while linking back directly to relevant timestamps for further exploration.

Study Design and Risk Factors in Craniosynostosis

Overview of the Study

• The discussion revolves around a cohort study aimed at identifying risk factors for developing craniosynostosis, initiated by the speaker during their medical career.

• The study was published internationally and involved reviewing clinical histories of 50 cases alongside 150 controls to establish a comprehensive understanding of the condition.

Methodology

• A cohort design was chosen due to the rarity of events like craniosynostosis, which necessitated a longer observation period rather than relying on immediate occurrences.

• The speaker emphasizes that when an event is infrequent, traditional cohort studies may not yield useful data; hence alternative designs are necessary.

Types of Study Designs Discussed

• Various study designs were mentioned: descriptive, prospective, retrospective, comparative longitudinal, and cross-sectional. Each serves different research purposes depending on the nature of the data being analyzed.

• The intention behind using a comparative approach is to evaluate differences between groups effectively.

Challenges in Data Interpretation

• The complexity arises when considering multiple occurrences of related conditions (e.g., urinary infections during pregnancy), complicating the analysis of risk factors for craniosynostosis.

• There’s an acknowledgment that while some questions posed in studies may be flawed or debatable, they still contribute valuable insights into understanding potential correlations.

Findings and Conclusions

• A case-control study conducted revealed no significant findings regarding risk factors associated with craniosynostosis despite thorough investigation into maternal health during pregnancy.

• The session concludes with plans for future discussions on bibliographic searches and encourages participants to bring laptops for practical learning.