Bio Statistics Core

GCRC Biostatisticians Susan M. Perkins, PhD 274-2626, [email protected]

Chandan Saha, PhD 278-4283, [email protected]

Our Role in the GCRC ♦ Project Support ♦ Protocol Review ♦ Follow-Up Review ♦ Teaching of Residents

Project Support ♦ Study Design ♦ Statistical Methods ♦ Sample Size and Power Calculations ♦ Data Management Advice ♦ Data Analysis

Protocol Review ♦ Hypothesis ♦ Study Design ♦ Data Collection ♦ Statistical Methods ♦ Sample Size and Power Calculations

A Research Project: Comparison of the Antihypertensive Efficacy of Torasemide and Furosemide in Patients with Chronic Renal Insufficiency ♦ Patients with chronic renal insufficiency show decreased

salt and water excretion which leads to systemic arterial hypertension. ♦ Torasemide and Furosemide have been found to be effective in increased salt and water excretion for these patients. ♦ Preliminary studies suggest that Torasemide has a bioavailability of 90-100%, but Furosemide has a bioavailability of 60%. ♦ This study aims at comparing these two drugs in Na excretion and antihypertensive effects.

GCRC Biostatistics Section ♦ Hypothesis: Primary and Secondary

● A statistical hypothesis is an assertion about a population. ● Example: Torasemide is superior to Furosemide in Na excretion and antihypertensive effects.

GCRC Biostatistics Section ♦ Study Design ● Design is the process or structure that produces valid data to answer the research question. ● Example: (Randomized, double blinded, crossover design) Study participants received both drugs either Torasemide first and Furosemide second or Furosemide first and Torasemide second with one week washout period.

GCRC Biostatistics Section ♦ Data Collection

● Description of how and what data will be

collected for answering the research question.

● Example: For each treatment (Torasemide and Furosemide), both baseline and 24-hour post treatment Na excretion and BP data were collected at each period.

GCRC Biostatistics Section ♦ Statistical Methods ● Statistical tools to test the hypothesis. ● Example: If there is no carry-over effect,

treatment effects can be compared using a paired t-test on the response variable, D = (YT2 – YT1) – (YF2 – YF1), where T and F indicate Torasemide and Furosemide, and 1 and 2 indicate time before and after treatment.

GCRC Biostatistics Section ♦ Sample Size and Power Calculations ● How many subjects are required to detect an effect of certain magnitude with power x% (usually 80% or 90%). ● Power (probability of detecting an effect if it exists). ● Example: Using a paired t-test at the 5% level of significance, 15 subjects would provide 80% power to detect an effect that is at least 0.81 times the standard deviation.

Writing Biostatistics Section ♦ Hypothesis

● Clearly state the primary and secondary (if any) hypotheses.

♦ Study Design

● Describe how the study will be conducted. ● Will it be a single/double blinded design? ● How the randomization will be performed. ● How the study subjects will be enrolled. ● When will subjects be seen? ● Include a timeline.

Writing Biostatistics Section ♦ Statistical Methods ● Mention what descriptive statistics will be used. ● Choose an appropriate statistical tool to test the hypothesis. ● Justify the choice of the statistical tool to test the hypothesis ● Include if any confidence interval for an estimate will be reported. ● State how the assumption(s) (if any) of the statistical method will be tested.

Writing Biostatistics Section ♦ Data Collection ● Describe when, how and what data will be collected. ● Do you have all relevant data to test the hypothesis and for other secondary analysis?

Writing Biostatistics Section ♦ Sample Size and Power Calculation ● Pilot study? If so, describe. ● Minimum detectable difference. ● Variability assumed or from previous similar study. ● Power (often 80%) ● Significance level (often 5%) ● 1 or 2 sided ● Test used (usually the same as in statistical methods) ● What sample size do you need?

Writing Biostatistics Section ♦ Drug Study or NIH Grant? ● Need elements described earlier ● Should be in the protocol or grant

Follow-Up Review ♦ Provide our feed back to the investigators. ♦ Ask for a meeting with the investigators if

further clarifications are needed. ♦ Suggest the modifications if required in study design, statistical methods and sample size calculation. ♦ Review the second time submitted protocols.

Teaching of GCRC Residents ♦ Provide only to the GCRC Residents ♦ Understanding and Application of Statistical

Concepts to Clinical Research ♦ Textbook: Norma and Streiner, Biostatistics, The Bare Essentials

♦ One to One Session, 6-8 Sessions, 30 – 60

Minutes Per Session ♦ Meet Twice a Week ♦ Provide Reading Assignments and Discuss the Materials