Home / Topics / Study Management / Data / Data Collection / Achieving and Maintaining Data Integrity
No Photo Available

Achieving and Maintaining Data Integrity

Posted on March 28, 2006

Linda Moody (bio) describes a method of maintaining data integrity during randomized clinical trials.

 

Q: Why is data integrity important?
A: The results of randomized clinical trials would be impossible to interpret without data that is valid, sufficient, and high-quality. Anything else would be a waste of researchers' and clinicians' time. Remember also that we have an ethical obligation to make good use of our participants' involvement in our studies, and that requires us to make every effort to ensure that our data meets the highest standards.
Q: What can I do to ensure data integrity in a randomized clinical trial?
A: When you're planning and conducting clinical trials, you need to develop procedures that will allow you to continually observe, monitor, and evaluate all aspects of the study. We developed a conceptual schema of data integrity based on NIH guidelines. The process of achieving and maintaining data integrity requires attention to methods of data collection, data collection training, and data monitoring.

Data collection
Develop your data collection protocol within the setting from which you will draw your sample. The research team and the clinical staff must work together to develop a protocol that is satisfactory to both. If data collection is burdensome, for example, clinical staff may "protect" the patients whom they feel would become tired or agitated by data collection procedures. Researchers, therefore, need to work with the staff to use measures that are not too arduous to the patients while still providing the needed data.

Random assignment of patients to treatment or control groups also needs to be discussed. The research team needs to communicate with the clinical staff so that everyone understands why the random assignment is important and how that fits in with the overall goals of the study. Assigning a participant to a less effective treatment would be unethical, of course, but random assignment to groups may be ethically justified if there is no consensus that one treatment is more effective than another.

Steps also need to be taken so that the research team is not seen by the clinical staff as interfering in the standard care of patients. In our study of hospice patients, for instance, we assured the staff that researchers would not attempt to answer questions about standard care and would instead refer patients back to the clinical staff. We also postponed initial data collection by 24-48 hours because the clinicians were concerned that data collection upon admission to the hospice, which is a distressing time, would be too burdensome to the patients.

Before you begin data collection, the data collection protocol needs to be clearly defined so that every participant will be treated consistently. This includes providing a distraction-free environment for data collection so that participants can focus on the questions and give accurate responses.

Also, be sure to pre-test the instruments to be used in the study. Conduct a pilot study with people who meet the study's inclusion criteria, and make an effort to get a diverse sample. The pretesting will allow you to give accurate time estimates to the clinical staff and the participants, and you may discover new ways to present the measurements to reduce the burden placed upon participants. As an example, our pilot study led us to develop large print flashcard versions of our measurements for patients too fatigued to sit up in order to complete the forms.

Data collection training
Once you've developed the data collection protocol, you need to think about how you will train the research staff to use it. For our study, we conducted a week-long training program that included role-playing and simulations. These can be very useful if you provide a variety of scenarios that the research personnel may encounter in the study and then provide constructive feedback. If appropriate, include 'crisis' scenarios so that the research personnel have the opportunity to develop appropriate strategies in a non-threatening environment. You may find your protocols will be revised and improved as a result of the training because you will learn more about the process as well.

Data monitoring
When you begin data collection, you need to have a data monitoring protocol in place. The way you monitor the data will be influenced by several factors such as the associated risks of the study and the trial's size and complexity. Think about how you will minimize missing data and how you will assess fidelity to the data collection protocol. Our study involved two data collectors per visit, and we had them exchange forms to check for completeness before they left the site. Our data entry staff also checked the records for missing data. In the event of missing data, the data collector was called in an attempt to recover the data, and we also gathered information about what caused the missing data, how we could prevent future missing data, and suggestions for revisions to the protocol. Even with the best protocol possible in place, there will still be missing data, and researchers have to decide in advance how they will handle that. Large multi-site trials are required to have data safety monitoring boards for ongoing evaluation of the data. Investigators in smaller trials typically develop a database for monitoring the data integrity and check for outliers in the data as well as well as consistency.

 

 


Based on published article and personal communication with the researcher in March 2006. Moody, L., & McMillan, S. (2002). Maintaining data integrity in randomized clinical trials. Nursing Research, 51(2), 129-133.

Feedback

Please note that the feedback is viewed only by 4researchers staff and is not intended for communication with individual contributors.

 

Use the form below to submit feedback about this article. If you would like a response, please be sure to include your e-mail address.


More About "Data Collection"

3:40

Ask Questions That Make Sense in the Culture

2:42

Asking Concrete Questions

Show All...