1 / 96

Planning Research, Collecting and Analyzing Research Data

Planning Research, Collecting and Analyzing Research Data. Health Insurance in Transition - 5th International Conference September 26, 2002, Zagreb, Croatia. Yen-Hong Kuo Jersey Shore Medical Center Meridian Health System Neptune, New Jersey U.S.A. Outline. Introduction Planning Research

Télécharger la présentation

Planning Research, Collecting and Analyzing Research Data

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Planning Research, Collecting and Analyzing Research Data Health Insurance in Transition - 5th International Conference September 26, 2002, Zagreb, Croatia Yen-Hong Kuo Jersey Shore Medical Center Meridian Health System Neptune, New Jersey U.S.A.

  2. Outline • Introduction • Planning Research • Collecting Research Data • Analyzing Research Data • Conclusion

  3. Medical Research Observe Phenomenon Propose Hypothesis Plan Research Collect and Analyze Data Interpret Results

  4. Example • Some of my friends are eating vitamin E to prevent colon cancer. • Hypothesis: Vitamin E can prevent colon cancer • Questions • How many people are eating vitamin E? How many people are having colon cancer? • What is the association between eating vitamin E and having colon cancer? • Can vitamin E prevent colon cancer?

  5. Purpose of Research • Description • Estimation • Making comparisons • Assessing association

  6. Planning Research • Literature Review • DEFINITION • Consult Experts • Study Design • Sampling Arrangement • Data Collection Arrangement • Statistical Analysis Plan • Practical Considerations

  7. Literature Review • What has been done and how • Population • Study design • Statistical methods • What has not been done • Limitation of the studies • Issues missed in the previous studies • Learn more about background info • Find more relevant studies

  8. DEFINITION • The key to a successful study • “A Problem well defined is a problem half solved” • Refine the hypothesis • Be specific • Define variables • Outcome variables • Explanatory variables • Biological meaningful and clinically useful

  9. DEFINITION (continued) • Clearly described in the protocol and report • Examples • Classification of BMI (body mass index) • Estrogen-Receptor Status (JAMA 2000;283:338-9)

  10. Consult Experts • When? • BEFORE design your study • Who? • Colleagues with experiences in study design and statistical analyses • Epidemiologist • Biostatistician

  11. Design of Analytical Studies • Observational studies • Cross-sectional / prevalence • Case-control / retrospective • Cohort / prospective • Experimental studies • Controlled clinical trials

  12. Cross-Sectional Study TIME No Disease Diseased Exposed Exposed Not Exposed Not Exposed

  13. Cross-Sectional Study (continued) • Data collected at a single point in time • A “snapshot” • Describe prevalence • Prevalence vs. Incidence • Assess associations

  14. Cross-Sectional Study (continued) • Strength • Quick • Cheap • Weakness • Can not establish cause-effect

  15. Case-Control Study Not Exposed Not Exposed Exposed Exposed Exposed TIME No Disease Diseased cases controls

  16. Case-Control Study (continued) • Starts with people who have disease • Matches them with people who do not have disease (control) • Looks history for exposures • Assesses associations

  17. Case-Control Study (continued) • Strength • Fast • Cheap • Useful to generate hypothesis • Good for rare diseases • Can examine several exposures • Estimates odds ratio

  18. Case-Control Study (continued) • Weakness • Can only study one outcome • Can not measure • Prevalence • Incidence • Relative risk • High susceptibility to bias

  19. Cohort Study Not Exposed Exposed disease free TIME Disease Free Develop Disease Disease Free Develop Disease

  20. Cohort Study • Starts with disease-free subjects • Classifies subjects as exposed or not exposed • Records outcomes • Assesses associations

  21. Cohort Study (continued) • Strength • Allows for accurate measurement of exposure variables • Estimates incidence • Estimates relative risk • Can measure multiple outcomes • Can adjust for confounding variables • Establishes time sequence for causality • Eliminates recall bias

  22. Cohort Study (continued) • Weakness • Time consuming • Expensive • Can not study rare outcomes • Confounding variables • Disease may have a long pre-clinical phase • Exposure may change over time • Attrition of study population

  23. Clinical Trial (therapeutic) Studied population (with disease) Randomized TIME Placebo Treatment No Disease No Disease Diseased Diseased

  24. Clinical Trial (preventive) Studied population (without disease) Randomized TIME Placebo Treatment No Disease No Disease Diseased Diseased

  25. Clinical Trial (continued) • Starts with subject with diseases (therapeutic) or without diseases (preventive) • Randomized • Blinding • Assignment • Assessment • Placebo controlled

  26. Clinical Trial (continued) • Strength • “Gold Standard” • Best design for controlling bias • Can measure multiple outcomes • Best measurement of causal relationship • Weakness • Expensive • Compliance • Ethical issues may be a problem

  27. Summary of Analytical Studies

  28. Meta-Analysis • Main Goal • Combine the results of previous studies to reach summary conclusions about a body of research • Steps • Identify studies with relevant data • Define inclusion and exclusion criteria • Abstract data • Analyze abstracted data statistically

  29. Sampling Arrangement • Population • Inclusion and exclusion criteria • Methods of Sampling (observational) • Sample Assignment (experimental) • Sample Size • Confounding • Bias

  30. Methods of Sampling • Probability Sampling • Simple random sampling • Systematic sampling • Stratified sampling • Cluster sampling • Multistage sampling • Multiphase sampling • Area sampling

  31. Methods of Sampling (continued) • Non-Probability Sampling • Subjective (Judgment) sampling • Convenient (Chunk) sampling • Quota Sampling

  32. Sample Assignment • Randomization • Fixed Allocation Randomization • Simple • Blocked • Stratified • Adaptive Randomization Procedures • Baseline adaptive randomization procedure • Response adaptive randomization

  33. Sample Size • Does size matter? • Why do we care? • Ethical Consideration • Time • $$$ • How to calculate? • Table • Software • Expert

  34. Sample Size Calculation • Clinically significant effect • Ho and HA • Type I Error • Type II Error (or Power) • Predicted rate of outcome events or Variance in the outcome measure

  35. Confounding • Confounding is an apparent association between disease and exposure caused by a third factor not taken into consideration • A confounder is a variable that is associated with the exposure and, independent of that exposure, is a risk factor for the disease

  36. Confounding (continued) • Examples • Study A found an association between cigar smoking and baldness. • What is a possible confounder? • Study B found an association between lead-paint ingestion and low IQ. • What is a possible confounder?

  37. How to Control Confounding • Study Design • Restrict study eligibility • Match on confounding factor • Analysis • Stratify on the confounding factor • Adjust for the confounding factor • Multivariate analysis

  38. Data Collection Arrangement • Design of data collection form or instrument • Instructions for data collection • Accurate / valid • Precise / reliable • Complete • Unaffected • Data storage and security • Confidentiality

  39. Statistical Analysis Plan • Statistical Methods • Software • Who is responsible for data analysis

  40. Practical Considerations • Budget • Time • Supports • Experiences • Ethics

  41. Collecting Research Data • Follow the Data Collection Arrangement • Coding for Missing Values • Data Arrangement for Statistical Analysis • Backup Data

  42. Coding for Missing Values • Example • Question: Does ID #002 have any symptom?

  43. Data Arrangement • Case I • Case II • Question: Which one is better?

  44. Analyzing Research Data • Purpose of Statistical Analyses • Commonly Uses Statistical Methods • Issues in Statistical Analyses • Common Statistical Errors

  45. Purpose of Statistical Analyses • Describe data • Estimate parameters • Make comparisons • Assess association

  46. Commonly Used Statistical Methods • Table IV. Statistics used • ___________________________________________________________________________ • Student’s t test 64 (44%) • 2 63 (43%) • Fisher’s exact test 30 (21%) • Analysis of variance 21 (14%) • Linear regression 17 (12%) • Mann-Whitney U test 11 ( 8%) • Logistic regression 11 ( 8%) • Wilcoxon rank-sum test 9 ( 6%) • American Journal of Obstetrics and Gynecology 1996;175:1138-41

  47. Commonly Used Statistical Methods (continued) • Describe data • Frequency and Frequency Distribution • Central Tendency • Mean vs. Median • Dispersion • Range vs. Interquartile Range • Variance vs. Standard Deviation • Estimate parameters • Confidence Interval

  48. Commonly Used Statistical Methods (continued) • Make comparisons • t-test vs. Wilcoxon Rank-Sum Test • ANOVA vs. Kruskal-Wallis Test • Chi-square test vs. Fisher’s Exact Test • Assess association • Relative Risk; Odds Ratio • Regression analysis • Correlation

  49. Example • Clinical trials were conducted to evaluate a new weight loss drug. • Outcome: Satisfaction Group Yes (%) n1 n2 Placebo 45 100 1000 Drug 55 100 1000 • Questions: Can this drug work?

  50. Hypothesis Testing • Research Hypothesis • The research hypothesis is the conjecture or supposition that motivates the research • Statistical Hypothesis • Statistical hypotheses are hypotheses that are stated in such a way that they may be evaluated by appropriate statistical techniques • null hypothesis (HO) and alternative hypothesis (HA or H1)

More Related