Biostatistics But why Why do we read scientific

Biostatistics

But why? • Why do we read scientific litterature? • How do we read scientific litterature?

Different types of literature • Experimental studies – Clinical trials • Observational studies – Cohort study • Longitudinal, and prospective, time and patient consuming – Case-control study • Can be applied with low sampling number, difficult to choose the control – Cross-sectional study / survey • Inexpensive, historical, provides the current – Case-series study • Usually reports unexpected clinical observations • Meta-analysis • Reviews

Anatomy of most articles • • • Abstract Introduction Methods Results Discussion Conclusion

Abstracts • Usually structured as a mini-article • The purpose is to ‘sell’ the article • By reading the abstract you should be able to tell the conclusions made by the authors • By only reading the abstract you cannot judge the validity of the conclusions. • Consider this: – If the study is well performed, is the results interesting? – If the results are statistically significant, is the magnitude of changes/differences clinically relevant? – If the results are not statistically significant, was the sample size large enough?

Introduction • Justification / Rationale, why was the study done? • Context • Hypothesis • Aim of the study – Usually the last paragraph of the introduction. • Population in the study – Location, time, subjects

Method I • How was the study performed – You should be able to understand the method section, and if you are familiar with the research field you should be able to reproduce the experiment • What is studies? – Patients, Which? Exclusion / inclusion • How is the study performed? • How is the data analyzed? – E. g. normalization • Which statistics is applied? – Are assumptions violated?

Metode II (Subjects) • How are the patients selected? – How are they randomized? • What are the inclusion/exclusion criterions? – Is it reasonable? – Are these patients relevant for your research? • Are there follow-up? – How are they handled? • How are withdrawers handled? – Intention to treat

Metode III (subjects)

Metode IV (subjects) • Bias by selection of patients – Prevalence • E. g. if patients die before inclusion

Method V (Subjects) • Bias by selection of patients – Admission rate bias • E. g. if control and treatment groups are not concurrent – Non-responders and voluntariness • Recall the example with vaccine from first lecture. – Grouping bias • Causality and the healthy worker – Choice of procedure • • Deductive procedures Different procedures for control and treatment groups Concurrency Randomization

Method VI • Procedure in the study – Procedure bias • If the groups are not treated in the same way • If e. g. one treatment group demands more follow-ups than other treatments – Memory bias • Use diary instead – Apparatus bias • Inaccurate devices • Poor handling – Diagnostic bias • E. g. if two groups are not diagnosed in the same way. – Compliance • It may be more difficult for patients/subjects to cooperate if one treatment is painful or unpleasant than others

Method VII • Measurements – Variations in the measurements • Whatever is measured will have a natural variation • The person who measures is not measuring accurately e. g. røntgen • Some outcomes are not well-defined, e. g. pain • The device or the method is not accurate or indirect – Reliability and validity • New methods must describe the usefulness – Blinding – Data quality, questioners, multi-center studies

Method VIII • Number of subjects – Type II error: No difference were found due to low number of subjects. – Power test, if negative results are reported

Method IX • Statistical methods – Are the methods appropriate and valid? – Are the assumptions violated? – Is it a fishing trip? – Multi-significance • Type I errors, false positives • Analysis along the way must be scheduled before performing the study – Migration bias • Intention-to-treat – Entry time

Results • This is where results, figures, tables and the statistics are presented • Does the results support the conclusion? – What is the baseline? – Risk of multiple comparison errors – Is the result consistent?

Diskussion og konklusion • Is there connection between the hypothesis, results and the conclusion? • Are there any short-comings? • How does the conclusion affect your research?

Opgaver • Discuss Leibovici (2000) • 7, 8, 9, 11, 12, 66, 67, 68, 69