JOHNS HOPKINS MEDICINE
Black spacer ABOUT JOHNS HOPKINS MEDICINE HEALTH INFORMATION PATIENT CARE RESEARCH EDUCATION Hopkins Medicine Home Page
image
scholar program abovenews
Biostatistical and Clinical Research Scientific Resources Core

blank bar
FAQ

What is the difference between Quality Assurance and Quality Control?

Quality assurance is all the planned and standardized activities implemented within the quality management system to provide support for data that will satisfy the defined quality standards in order to limit the introduction of errors and yield data of adequate quality and usability for their intended purpose.

Quality control involves monitoring specific data to determine if they conform to the defined quality standards, and identifying ways to reduce, eliminate, or prevent deficiencies that yield unsatisfactory results.

In other words, QA is essentially the documented plan implemented to ensure quality data acquisition. QC is the process of looking for, detecting, and fixing any defects in the data that affect quality.

What do I need to provide for a sample size calculation or power analyses?

Sample sizes are an important part of the design of your research study. Basically, this allows you to determine if the number of patients you can logistically recruit for your study will provide the evidence necessary to support your conclusions. Put in statistical terms: given that the alternative hypothesis is true, how many patients are needed to have adequate probability (e.g. power) to detect a statistical difference? Ideally, the desired power is on the order of between 80 and 90 percent.

In order to perform a sample size calculation, you need two basic pieces of information: 1) some idea of the effect size (e.g. difference) that you expect to see between the treatment and control groups, and 2) some estimate of the variability of this treatment effect. In most cases, the effect size is determined from pilot data. When pilot data are not available, published studies with similar data might provide an estimate of the effect size you expect in your study.

Formulas for simple sample sizes based on t-tests or differences in binary proportions are given in Fundamentals of Biostatistics by Bernard Rosner1, while sample size equations for more complicated studies are given in Sample Size Calculations in Clinical Research by Shein-Chung Chow, Jun Shao, and Hansheng Wang2. In addition, several sample size calculators are available on the internet that tend to be user friendly.

References: Rosner, B. (2000). Fundamentals of Biostatistics, 5th Edition. (Duxbury: Pacific Grove, CA).
Chow, S.C., Shao, J., and Wang, H. (2003). Sample Size Calculations in Clinical Research. (New York: Marcel Dekker, Inc.).

Why Data Dictionary?

A data dictionary is a "codebook" to understand the meaning of the collected data. For each variable, the data dictionary should include: a). type (eg, continuous, integer, text, other) b). format (eg, "Yes", "No", "Missing") and c). permissible values (eg, date field can only include dates after January 1, 2005, or a response coded as 0, 1, or 99).

Most statistical rountines require that non-numeric information be coded into numeric answers. Coding is both an act of translation and an act of summarization. These numbers then become the values in a field of the electronic data file eventually produced.

It is best to code "Don't know" differently from missing data. In the analysis, all these responses are treated as missing, but the reason the data is missing is retained.

References:

Wayne Enanoria, PhD, MPH (UC Berkeley School of Public Health)

 

BCRSRC | © Copyright 2006 | All Rights Reserved