Thomas B. Newman, MD, MPH Professor of Epidemiology & Biostatistics and Pediatrics, UCSF

1. Epi 202: Designing Clinical ResearchIntroduction to the Course, Clinical Research and Research Questions Thomas B. Newman, MD, MPH Professor of Epidemiology & Biostatistics and Pediatrics, UCSF July 31, 2012

2. Outline • About this course • Anatomy and Physiology of Research • Research questions • Examples: jaundice in newborns

3. About This Course • Began > 30 years ago • Also known as the "Hulley Course" • Steve was the leader for the text (DCR) and designed the course, homework, and instructions to section leaders Steve Hulley

4. About This Course • Michael Kohn • Co-director, 2004 – 2011 • Joel Simon • Section leader, 1991– • Course co-director, 2012

5. Website • Google “Epi 202” or find from TICR home page • Course roster, schedule, rooms, readings, PowerPoint files (when available) • Links to recordings of lectures • This class • Epi 150.03/CTSI site • Forum

6. About the Reading -1 • DCR-3 includes exercises and answers at the end of the book • We recommend jotting down answers before reading ours • Can discuss in section but usually won’ t be turned in • Let us know your suggestions for improving the book! (Now working on DCR-4!)

7. About the Reading -2 • Recommended reading this week (Saha et al. Survival guide…) on the Epi 202 website • Evidence-Based Diagnosis (EBD) text also recommended; you’ll need it for Epi 204

8. Course Objectives 1. Learn about how to design and do clinical research 2. Produce a protocol for a study 3. Help others in the workshop 4. Provide feedback on the workshop 5. Have a multiplier effect

9. Course Ingredients Prefer 9:00 start? July 31- Lectures (9:10 – 10:00) Sept 11 * Selected issues from DCR 3 text and examples Sections (10:10 – 12:00) * Protocol components * More issues from the text * Helping and getting to know your classmates Sept 18 5-page protocols due Oct 2, 9, tba Protocol review sessions (not Masters or ATCR Students) *In pairs and trios, new faculty

10. Grades, Attendance, Interruptions, etc. • 5-page protocol must be turned in to pass • Protocol review sessions encouraged but not req’d • To get an A, in addition: • No unexcused absences • No more than 1 missing or late HW • Help your colleagues with their protocols • Class time is sacred • Do not answer cell phones or pages except emergencies • If clinical responsibilities will interfere, please arrange coverage or take this class another time

11. Faculty for sections *E-sections

12. Course Coordinator Olivia De Leon Olivia@epi.ucsf.edu 514-8231 (tel) 514-8150 (fax) (Please let her know if your email address changes by sending her an email from the new address) Olivia De Leon

13. Questions? USE MICROPHONE OR REPEAT

14. Anatomy of research: What it’s made of • Research question, significance • Study design • Study subjects and how they will be sampled • Variables and how they will be measured • Predictor • Outcome • Analysis plan, sample size calculation • Implementation, data management, quality control

15. Required Viewing for Next Week

16. Research Questions for Epi 202 • Not the best choice for this course • Animals, molecules without humans • Data syntheses, e.g. decision analysis, cost-effectiveness analysis, meta-analysis • Qualitative research • Ideal • A new observational study or clinical trial involving humans that you could do (or at least start) this year

17. What if I am doing a secondary data analysis? You can: • Use it for your DCR project, rethinking decisions that were already made and getting thoughts and suggestions for colleagues • Design a new study you may not be planning to do

18. Physiology of research: How it works Using measurements in a sample to draw inferences about phenomena in a population

19. DCR Figure 1.3

20. DCR Figure 1.4

21. DCR Figure 1.5

22. Questions? USE MICROPHONE OR REPEAT

23. Do I really have to do all of those laboratory tests before I can start phototherapy in jaundiced babies? Newman research question #1

24. Background about neonatal jaundice • Bilirubin: Yellow breakdown product of heme (from red blood cells) • Jaundice: Yellow color of whites of eyes and skin due to high bilirubin. Usually indicates liver or blood disease, but generally is normal in newborns

25. Background to Question #1, cont’d • Phototherapy: shining light on the baby’s skin helps lower bilirubin levels • Very high bilirubin levels (“hyperbilirubinemia”) can cause kernicterus (brain damage) • More common in developing countries

26. Do I really have to do all of those laboratory tests before I can start phototherapy in jaundiced babies? My first grant: Laboratory Evaluation of Jaundice in Newborns (“LEJN”) Newman research question #1

27. Digression: the importance of a good acronym • Fun initial way to engage collaborators • Gives your study credibility and life • Useful for naming files and directories • Worth the effort!

28. The importance of a good acronym: Examples • Multiple Risk Factor Intervention Trial • = MRFIT • Late Impact of Getting Hyperbilirubinemia or photoTherapy • = LIGHT

29. The importance of a good acronym: Examples • A study of the cost effectiveness of differerent protocols for treating gestational diabetes (James G. Kahn, MD, MPH) • Gestational Diabetes Formulas for Cost-Effectiveness GeDi FORCE

30. Background to Question #1, cont’d • A complete "hyperbilirubinemia work-up" used to be recommended for significant newborn jaundice: • Total and direct bilirubin • Direct and indirect Coombs’ tests • Complete Blood Count • Blood smear for red cell morphology • Reticulocyte count • Urine reducing substance

31. Background to Question #1, cont’d • In TN’s experience reference ranges were poorly defined and results rarely if ever affected management • As a pediatric resident TN did not like having to get out of bed to draw blood for these tests before being allowed to start phototherapy

32. TN concerned about costs

33. More refined research question #1(i.e., what we really want to know) Do the expected health benefits of the recommended tests justify their costs? • Subjects: Jaundiced newborns (candidates for phototherapy) • Predictor variable: obtaining the tests • Outcome variable: measurements of health and costs

34. Laboratory Evaluation of Jaundice in Newborns (LEJN) study questions (i.e., questions our study can answer) • How often are each of these tests done in newborns at UCSF and Stanford? • How often are they abnormal? • When they are abnormal what diagnoses are made as a result of the test? • How often is treatment altered? • Diagnostic yield study (Chapter 12); primarily descriptive

35. Compromises • Just 2 S.F. Bay Area teaching hospitals • Surrogate outcome: • Discharge diagnosis of a significant disease • Diagnosed after an abnormal jaundice work-up • Retrospective study • Limited to those in whom MD ordered the tests, rather than those with a certain level of jaundice or meeting other inclusion criteria • No control over laboratory methods for doing the tests

36. Is RQ /Study Plan FINER? Feasible Interesting Novel Ethical Relevant

37. Can you put your FINGER on a good research question? Feasible Interesting Novel Good for your career Ethical Relevant

38. Good for your career Try to identify a research question that will allow you to • Learn more about an area of potential long-term interest • Acquire new skills you could use on other projects • Work with people and/or organizations with whom you want to develop a long term relationship • Build on the project for future work

39. LEJN: Direct Bilirubin Results -1 • Test ordered 15 times as often per infant at UCSF as at Stanford • Results more than twice as high • Total N births • Stanford: 5,185 • UCSF: 5,778 1 2 3 4 mg/dL AJDC 1991;145:1305-1309

40. LEJN Results: Direct Bilirubin Results -2 AJDC 1991;145:1305-09 Spontaneous resolution in all 4 infants

41. LEJN Conclusions • “Because of their low yield and poor specificity, direct bilirubin tests are seldom helpful in evaluating jaundice in term newborns.”* • Current guidelines: check direct bilirubin for unexplained rapid rise, babies needing phototherapy, persistence > 3 wks or sick baby *AJDC 1991;145:1305-1309

42. Background to TN RQ #2 • It is known that very high (> ~30 mg/dL) bilirubin levels can cause horrible brain damage (kernicterus) • Unclear how often kernicterus or more subtle abnormalities occur at lower bilirubin levels • Concern about this possibility leads to more treatment • Bilirubin levels  25 mg/dL are rare (~1/700)

43. Background to TN RQ#2, cont’d • During the 1990s hospital stays for newborns shortened dramatically • There were reports of increases in kernicterus and severe dehydration • We had already identified cases of bilirubin  25 mg/dL and dehydation from previous nested case-control studies • RQ: What are the effects of neonatal bilirubin levels 25 mg/dL and dehydration on neurodevelopmental outcomes?

44. Acronyms • Sequelae of Hyperbilirubinemia and Dehydration in Infants • “SHADI” • Jaundice and Infant FEEding Study • JIFee

45. Study Design • Triple Cohort Study • Hyperbilirubinemia group (TSB  25 mg/dL at < 30 days) • Dehydration group (readmitted for dehydration + either  12% weight loss or Na >= 150 mEq/L) • Randomly selected comparison group • Outcomes: blinded full neurodevelopmental evaluations at age ~5 by psychologists and child neurologists

46. Compromises and challenge • Difficulty recruiting “controls” • Full exams on 82/140 (59%) hyperbili cases vs 168/419 (40%) of controls • Outcomes • Interobserver variability, subjectivity in examinations • Measurements at age 5 years may miss relevant school problems later • 5-year-olds get tired and have bad days • No hearing tests

47. Results: continuous variables

48. Results: Adjusted OR and 95% CI for Dichotomized Outcomes

49. Publication • Submitted to JAMA • Rejected • Submitted to NEJM • Rejected • Lower participation rate in controls (40% vs 59%) • Questionable importance

50. Decision appealed! • Even if all unexamined controls normal, no change in results • Google search • Timely e-mail