Dates: the silent death in medical device clinical trials

Bad Dates: Assessing and assuring high quality dates in clinical trials

Introduction

Clinical trials are based on collections of time-based clinical data. If the dates and time-stamps in the data set are low quality, everything else will be low quality: measurement of study progress, enforcement of visit protocols and study schedules, measurement of site progress and any clinical parameter that is a function of time, such as cumulative dosing, pregnancy and hundreds of other time-based use cases.

Jenya talks about bad dates and how really bad quality dates that can spell disaster for your clinical trial – and suggests what do to about it.

I think everyone knows that secure and accurate clinical trial data, no matter where and what data, should be correct, well written, accurate and qualified. Without that, no one will trust your data, your results and all your hard work will be non-appreciated by others and too frustrated to you. Today I’d like us to be more specific and focus on date quality in clinical trials. Dates in clinical trials are used in every visit, in any form, and across changes. I think that we can all agree that your clinical data will be full of date and time information.

Do we take date quality for granted?

Dates are important, but maybe this importance is taken for granted? In our work in remote monitoring of data in clinical trials, I’ve discovered some interesting insights regarding date quality. I’ve seen dates that were entered in shall we say, some very very creative ways. These examples underscore the need for people to know the importance of date fields in clinical trials but unfortunately we see cases where sites don’t pay enough attention on how and when they collect dates and disregard the consequences of their creativity.

The importance of date information for remote monitoring.

It’s not a secret that I’m a big fan of clinical trial monitoring automation (our team is working hard on developing a great tool for remote monitoring of clinical trials). One of our main goals is to help sponsors bring studies to completion with less (and hopefully zero) background noise. The objective should be to work with less background noise throughout the trial and not wait until the end in order to clean up just before statistical analysis and regulatory submissions.

Date/time issues can be visualized as a triangle.

The first side of the triangle is the CRC who collects data into an EDC system; the second side of the triangle is the CRA who monitors CRC work and data quality and performs SDV. The third side of this triangle is the subject who needs to come and visit the doctor on certain days that CRC assigned him when he started the trial.

Using remote monitoring we can see if CRC and CRA are working according the protocol and subject sees his doctor on time. If you ask yourself what I’m talking about, you will discover it by the end of this paragraph. In particular, every CRC needs to type all subject’s information regarding every visit during or immediately after it. Things can happen (we are all people) and visit information can be data entered within a day or 2, after the visit which is usually fine. But the worst cases is when this kind of information is typed a month or 2 or 3 later or entered in advance of the date of visit. Another bad scenario is not recording adverse events on time and informing the right people. This is wrong, unprofessional and of course, a protocol violation; by using the remote monitoring tools this inadequacy may be prevented before further damage happens during the trial.

Another example is CRA performance on sites, how they check CRC’s work, how often they do SDV, how they react to queries and close them. Every visit should be considered in advance and of course, performed on time. CRA’s progress is very important and a small deviation in dates on their visits may cause some gaps on sites progressing in the study. By using remote monitoring, it is possible to follow CRA’s progress and make sure that data entry and SDV are done on a timely basis.

Subject’s participation on time in the study is very important, or the most important, to data quality. Every delay in a visit is a protocol violation that must be recorded and may result in early termination from the study. Writing a wrong date, even by mistake, and have a query on it will not resolve the problem, it can make the situation worsen. Early termination of small amount of subjects can be tolerated, and I’m pretty sure this is taken in account but what can happen if there are many cases of early termination? Maybe there are problems in the study and during a big multi-center trial it is hard for the clinical operations team to zero in or even report issues to the right people. Remote monitoring can help the clinical operations team deal with that kind of problem! It is better to pause the study, have second thoughts and maybe do some amendments, rather than complete the study with unprofessional and unqualified data.

Never underestimate the capability of people to make mistakes

As I mentioned before, we discovered some very creative dates formats and I think it’s about time to share with you some of them. Are you ready for that?

Generally, cloud-based Web applications for remote monitoring of patients in clinical trials collect date information in a standard dd-Mmm-yyyy format. using date pickers – standard HTML/JavaScript calendar widgets designed specifically for the purpose of ensuring capture of valid dates. You just need to click on a calendar widget and select your required date and ensure recording of a valid date. Sounds easy, no? However, we keep discovering how creative people can be. Instead of clicking inside a calendar widget, they manually enter inventive new dates.

Here are some examples of user creativity in typing that defeat ECRFs: typing Roman numerals (01-XI-2015), typing dates in the local languages of the team  (01-noi-2015), typing unknown date codes (uk-unn-2015) or combinations of the above (nk-XI-2015). There is no limit to creativity in date data entry.  To make matters worse, we’ve often observed that in many of these cases, queries were fired and closed by the CRA because they paid more attention on data existence than to data quality.

The consequences of low quality dates in clinical trials

Clinical trials are based on collections of time-based clinical data. If the dates and time-stamps in the data set are low quality, everything else will be low quality: measurement of study progress using outcome definition and measurement, enforcement of visit protocols and study schedules, measurement of site progress and any clinical parameter that is a function of time, such as cumulative dosing, pregnancy and hundreds of other time-based use cases. I’ll say it again – poor-quality dates in your clinical trial data can be disastrous.

How to resolve date quality issues?

There are several ways to reduce date quality issues, but I think the most important thing is to properly understand the importance of date data and the bad influence of bad data to the clinical trials. The first thing must be better TRAINING of site and CRA teams. They should know and remember that dates must be written in same formats during the trial. Moreover, all data must be entered into the EDC system as soon as possible, and in any case not before, in order to reduce any human errors. Another consideration is the EDC design; a well-designed study build will implement robust date validation at field level including cross-form validation (date of visit 2 must be at least 30 days after visit 1, etc).

And the last but not least we can ensure good quality dates by using remote monitoring tools (and I’m not biased!) that give clinical trial operations a clear and current picture of their date fields quality at all times; and believe me, remote monitoring can rescue your trial!

 

Free Ebook - Best practicies for implementing EDC Systems In Clinical Trials

Jenya Konikov-Rozenman

Jenya is FlaskData.io Customer Success Manager. Jenya has a masters degree in biotechnology from the Hebrew University and is a doctoral candidate at Tel Aviv in medical science. She is GCP and CRA certified and leads FlaskData.io customer operations with super-human devotion to customer delivery. Jenya dances and skis in her spare time.